Pianotech

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Thanks for doing this, Anthony.  It clearly took a lot of time to put this together.

One question:

At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

I have a PART 2 video to The Piano Sound Is a Moving Target. I posted this as a separate thread but somehow it didn't show. 

https://youtu.be/KIxAY4u__gQ

Here is the first review on YouTube From Rick Clark

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 01:36
From: Steven Rosenthal
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

I notice you are still arbitrarily defining the offset as "zero" for all the partials in the first time window. 

Anyway, that's not what I was coming here to say. I actually wanted to respond to your implication that I'm "bending" the data in my analysis due to bias. I'm not sure why I would want to bend data, or what I would be trying to "bend" it to show. The data is what it is, regardless of how you analyze it. So in an effort at transparency, I'm making the 8 recordings I analyzed publicly available at the following link: Steinway D, 2 notes, 4 sensors. You are welcome to run these recordings through your own script and publish your critique. Or we could compare notes and maybe learn something from each other if your analysis comes out different than mine. If I made a mistake in my analysis I'd be grateful to have it pointed out to me. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 10:55
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I have a PART 2 video to The Piano Sound Is a Moving Target. I posted this as a separate thread but somehow it didn't show. 

https://youtu.be/KIxAY4u__gQ

Here is the first review on YouTube From Rick Clark

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 01:36
From: Steven Rosenthal
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Anthony,

I 'stand by my original findings' and the video is here:

https://youtu.be/LPG32Kr1nKM

I also show the invariance of the sensor to placement. That video is here:

https://youtu.be/xUfOnQWWoD4

I also stand by the 'fact' that the sensor is 'invariant' to this because it only looks at string movement. Acoustic (mic) variance in frequency comes from the thousands of eigenmodes of the soundboard which tilt the spectral peaks. I suggest you get a better resolution of the spectral peaks intra-bin. Again, the sensor only looks at string movement, not acoustic waves coming off the soundboard.

I don't need to analyze someone else's data when I have the facts of the physics and proof of my own experiment.

If your experiments are showing to the contrary then I can 'assure you' that there is something fundamentally wrong with your analysis. I am confident, fully confident in my findings. It was independently vetted my partner, MIT-PhD Prof. emeritus, who did the experiment himself and came to the same findings.

Given all the above, why would I be spending time debunking another finding?

Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 11:57
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I notice you are still arbitrarily defining the offset as "zero" for all the partials in the first time window. 

Anyway, that's not what I was coming here to say. I actually wanted to respond to your implication that I'm "bending" the data in my analysis due to bias. I'm not sure why I would want to bend data, or what I would be trying to "bend" it to show. The data is what it is, regardless of how you analyze it. So in an effort at transparency, I'm making the 8 recordings I analyzed publicly available at the following link: Steinway D, 2 notes, 4 sensors. You are welcome to run these recordings through your own script and publish your critique. Or we could compare notes and maybe learn something from each other if your analysis comes out different than mine. If I made a mistake in my analysis I'd be grateful to have it pointed out to me. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 10:55
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I have a PART 2 video to The Piano Sound Is a Moving Target. I posted this as a separate thread but somehow it didn't show. 

https://youtu.be/KIxAY4u__gQ

Here is the first review on YouTube From Rick Clark

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 01:36
From: Steven Rosenthal
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Anthony,

I 'stand by my original findings' and the video is here:

https://youtu.be/LPG32Kr1nKM

I also show the invariance of the sensor to placement. That video is here:

https://youtu.be/xUfOnQWWoD4

I also stand by the 'fact' that the sensor is 'invariant' to this because it only looks at string movement. Acoustic (mic) variance in frequency comes from the thousands of eigenmodes of the soundboard which tilt the spectral peaks. I suggest you get a better resolution of the spectral peaks intra-bin. Again, the sensor only looks at string movement, not acoustic waves coming off the soundboard.

I don't need to analyze someone else's data when I have the facts of the physics and proof of my own experiment.

If your experiments are showing to the contrary then I can 'assure you' that there is something fundamentally wrong with your analysis. I am confident, fully confident in my findings. It was independently vetted my partner, MIT-PhD Prof. emeritus, who did the experiment himself and came to the same findings.

Given all the above, why would I be spending time debunking another finding?

Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 11:57
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I notice you are still arbitrarily defining the offset as "zero" for all the partials in the first time window. 

Anyway, that's not what I was coming here to say. I actually wanted to respond to your implication that I'm "bending" the data in my analysis due to bias. I'm not sure why I would want to bend data, or what I would be trying to "bend" it to show. The data is what it is, regardless of how you analyze it. So in an effort at transparency, I'm making the 8 recordings I analyzed publicly available at the following link: Steinway D, 2 notes, 4 sensors. You are welcome to run these recordings through your own script and publish your critique. Or we could compare notes and maybe learn something from each other if your analysis comes out different than mine. If I made a mistake in my analysis I'd be grateful to have it pointed out to me. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 10:55
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I have a PART 2 video to The Piano Sound Is a Moving Target. I posted this as a separate thread but somehow it didn't show. 

https://youtu.be/KIxAY4u__gQ

Here is the first review on YouTube From Rick Clark

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 01:36
From: Steven Rosenthal
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Three questions: 

1. I noticed that these new graphs show zeros for the first two time windows instead of just the first window.
   Am I correct in assuming that you added zeros to the front end of your data, and not that all the partials in both the first and second time windows happened to have the same frequency? 
2. Ok, so you don't want to waste time analyzing the data I uploaded. Are you willing to upload the sound files you analyzed so I can run them through my Matlab script? I think it would be helpful to know if we're comparing apples to apples here. 
3. I was hoping you could clarify what you wrote above: "I also show the invariance of the sensor to placement." I remember you saying at other times that the sensor must be placed near the agraffe or the bridge termination, otherwise it gives inaccurate results. (This was in the context of you saying that Loren Kelley's test was invalid because he wasn't consistent in where he placed the sensor.)

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 12:37
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Anthony,

I 'stand by my original findings' and the video is here:

https://youtu.be/LPG32Kr1nKM

I also show the invariance of the sensor to placement. That video is here:

https://youtu.be/xUfOnQWWoD4

I also stand by the 'fact' that the sensor is 'invariant' to this because it only looks at string movement. Acoustic (mic) variance in frequency comes from the thousands of eigenmodes of the soundboard which tilt the spectral peaks. I suggest you get a better resolution of the spectral peaks intra-bin. Again, the sensor only looks at string movement, not acoustic waves coming off the soundboard.

I don't need to analyze someone else's data when I have the facts of the physics and proof of my own experiment.

If your experiments are showing to the contrary then I can 'assure you' that there is something fundamentally wrong with your analysis. I am confident, fully confident in my findings. It was independently vetted my partner, MIT-PhD Prof. emeritus, who did the experiment himself and came to the same findings.

Given all the above, why would I be spending time debunking another finding?

Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 11:57
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I notice you are still arbitrarily defining the offset as "zero" for all the partials in the first time window. 

Anyway, that's not what I was coming here to say. I actually wanted to respond to your implication that I'm "bending" the data in my analysis due to bias. I'm not sure why I would want to bend data, or what I would be trying to "bend" it to show. The data is what it is, regardless of how you analyze it. So in an effort at transparency, I'm making the 8 recordings I analyzed publicly available at the following link: Steinway D, 2 notes, 4 sensors. You are welcome to run these recordings through your own script and publish your critique. Or we could compare notes and maybe learn something from each other if your analysis comes out different than mine. If I made a mistake in my analysis I'd be grateful to have it pointed out to me. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 10:55
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I have a PART 2 video to The Piano Sound Is a Moving Target. I posted this as a separate thread but somehow it didn't show. 

https://youtu.be/KIxAY4u__gQ

Here is the first review on YouTube From Rick Clark

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 01:36
From: Steven Rosenthal
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Three questions: 

1. I noticed that these new graphs show zeros for the first two time windows instead of just the first window.
   Am I correct in assuming that you added zeros to the front end of your data, and not that all the partials in both the first and second time windows happened to have the same frequency? 
2. Ok, so you don't want to waste time analyzing the data I uploaded. Are you willing to upload the sound files you analyzed so I can run them through my Matlab script? I think it would be helpful to know if we're comparing apples to apples here. 
3. I was hoping you could clarify what you wrote above: "I also show the invariance of the sensor to placement." I remember you saying at other times that the sensor must be placed near the agraffe or the bridge termination, otherwise it gives inaccurate results. (This was in the context of you saying that Loren Kelley's test was invalid because he wasn't consistent in where he placed the sensor.)

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 12:37
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Anthony,

I 'stand by my original findings' and the video is here:

https://youtu.be/LPG32Kr1nKM

I also show the invariance of the sensor to placement. That video is here:

https://youtu.be/xUfOnQWWoD4

I also stand by the 'fact' that the sensor is 'invariant' to this because it only looks at string movement. Acoustic (mic) variance in frequency comes from the thousands of eigenmodes of the soundboard which tilt the spectral peaks. I suggest you get a better resolution of the spectral peaks intra-bin. Again, the sensor only looks at string movement, not acoustic waves coming off the soundboard.

I don't need to analyze someone else's data when I have the facts of the physics and proof of my own experiment.

If your experiments are showing to the contrary then I can 'assure you' that there is something fundamentally wrong with your analysis. I am confident, fully confident in my findings. It was independently vetted my partner, MIT-PhD Prof. emeritus, who did the experiment himself and came to the same findings.

Given all the above, why would I be spending time debunking another finding?

Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 11:57
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I notice you are still arbitrarily defining the offset as "zero" for all the partials in the first time window. 

Anyway, that's not what I was coming here to say. I actually wanted to respond to your implication that I'm "bending" the data in my analysis due to bias. I'm not sure why I would want to bend data, or what I would be trying to "bend" it to show. The data is what it is, regardless of how you analyze it. So in an effort at transparency, I'm making the 8 recordings I analyzed publicly available at the following link: Steinway D, 2 notes, 4 sensors. You are welcome to run these recordings through your own script and publish your critique. Or we could compare notes and maybe learn something from each other if your analysis comes out different than mine. If I made a mistake in my analysis I'd be grateful to have it pointed out to me. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-28-2024 10:55
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I have a PART 2 video to The Piano Sound Is a Moving Target. I posted this as a separate thread but somehow it didn't show. 

https://youtu.be/KIxAY4u__gQ

Here is the first review on YouTube From Rick Clark

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-28-2024 01:36
From: Steven Rosenthal
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Steve, thanks for the clarification, I was making a false assumption. So what we are seeing on the graph is all 3 strings, all the partials are perfectly in tune and stay that way, in phase, throughout the duration of the period shown. Is that correct or is there some sort of averaging involved in the analysis?

------------------------------
Steven Rosenthal RPT
Honolulu HI
(808) 521-7129

Original Message:
Sent: 06-27-2024 19:00
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Nathan, all my published measurement unless otherwise noted, are fully coupled unmuted unisons. The single strings do not exhibit the Weinrich characteristic but look more like simple exponential decays. --- Steve N.

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 18:45
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

IIUC these are single string measurements, so if you're talking about the bumps and dips in the fundamental around the 1-1.5s mark, I don't think those have anything to do with the unison coupling effect from Weinreich.  It's probably more of a mic measurement artifact from reflections, phase cancellation, or something to that effect.  My own take-away is that when using a mic (for better or worse) I may need to either use the reading from right after the attack, or wait a couple seconds for the pitch to settle out.  Or, as I only just realized, try a higher partial.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 18:18
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

This is looking better, though I'm having some difficulty interpreting the last set of graphs. Nonetheless, AFAICS this all simply confirms what Gabriel Weinrich wrote about back in 1979 as well as what Virgil Smith wrote about much later, not to mention what Steve N. has been writing and talking about. 

Personally I'd like to see more cooperation in this endeavor as opposed to competition. If you guys could all put your heads together on it I think it would be more beneficial for all of us. It seems to me that the tools for recording and the software for analyzing is better than ever now. 

As to defining the best spot for determining unison tuning, my position is that technically speaking the first half second would be optimal, however if the resulting unison sounded stale or sterile or short or whatever as a result, I would try tweaking it a little to see if I could improve it. If I could not, I'd put it back there and move on. 

My mentor taught me how to do it in an analog procedure. He could do it extremely well.  I believe there is room at the table for both dead on unisons and slightly spread unisons. 

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 16:34
From: Steven Norsworthy
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Indeed, the fundamental is of course what we hear the most, and it has a very significant Weinrich Drift off the attack. Let's all make sure we are using all 3 strings in our graphs, unmuted.

In one sense, what is the point in just showing the mean of the changes of the partials if we are hearing the fundamental the most and it is the wildest one fluctuating?

If we show the standard deviation of the changes in the partials, it is significant. From my experiments the standard deviation is at least 1 cent from time slot to time slot. 

As I stated in my original video, what is the basis of the piano tuner's choice of picking a time window? I addressed that rhetorical question with answers in my video. Go back and review what I said. It has to do with both the power level (highest in the first second after the attack) and the music aspects of the nature of the piano musical literature. Now, I would revise and add to that the importance of the fundamental and our perception of it.

Steve

------------------------------
Steven Norsworthy
PianoSens
Cardiff By The Sea CA
(619) 964-0101

Original Message:
Sent: 06-27-2024 16:31
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Looking at the graphs you just posted, I need to retry that experiment and select a higher partial.  Seems like it's mainly the fundamental that goes wonky.

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Follow-up question on this one for Anthony: In your original video you show the results of averaging the partials.  It's been a hot minute since I took statistics but, aren't there other methods that are better at reducing the influence of statistical outliers, that might give an even more stable reading?  I realize you might have just used average in the video for simplicity's sake.

I suppose that only applies if the deviations at any given partial are independent of one another.  Looking at the graphs, I almost want to say that an noticeable dip in one of the partials almost seems to be correlated with a slight rise in the others.  But I could be overthinking it...

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

OK, so let me clarify. 

When I said I was using the average, I was basically taking the average frequency of each partial over the entire 3-second piano note. I did this for each individual partial from each individual sensor (3 mics and one pickup). 

Here's the actual data from the C5 note. 

After I had an average frequency reading from each sensor, I averaged the four frequencies from the four sensors together to get a single average frequency for each partial. Those are the numbers on the right. You can see that the measured frequencies are all really close to each other (standard deviation under 0.2 Hz even for the worst one). 

Then I went back to the original "partials over time" data and calculated the cents offset for each time window relative to the 4 "average" frequencies in the table above. So each partial is being measured and analyzed independent of the others, and they're all being plotted on a consistent scale so you can directly compare one graph to the others. When you see one partial go up and 3 others go down, it's because that's just what they happened to do, and not a result of how I analyzed the data. 

The point of this exercise wasn't to find a way to get a stable reading. It was to demonstrate that there are better ways to graph the data than just using whatever random frequencies happened to show up in the first time window (hammer strike to 0.5 seconds) as the standard against which you're plotting all the following dots. 

There is a better way still to plot these that I didn't do, but that I plan to at some point in the future when I find the time. It is to calculate target frequencies for the partials based on an inharmonicity model of the string and plot the deviation over time from those the targets. That's what Frank Illenberger was doing in the video he uploaded on PianoWorld. (Here's a link to that: https://www.youtube.com/watch?v=lWZXgrvFolU ) 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-30-2024 17:49
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Follow-up question on this one for Anthony: In your original video you show the results of averaging the partials.  It's been a hot minute since I took statistics but, aren't there other methods that are better at reducing the influence of statistical outliers, that might give an even more stable reading?  I realize you might have just used average in the video for simplicity's sake.

I suppose that only applies if the deviations at any given partial are independent of one another.  Looking at the graphs, I almost want to say that an noticeable dip in one of the partials almost seems to be correlated with a slight rise in the others.  But I could be overthinking it...

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Apologies for any confusion -- I understand the _main_ point of the video is to show a way of graphing the deviation of the partials from a meaningful zero point, i.e. without randomly biasing them based on a single time window.  If I got you properly just now, you explained that there were a couple steps of averaging that you used to do that in the first place.  Appreciate the additional info :)

But if I understood the video right, there's a third type of average that you illustrate with a thick line around 17:45 in the video, that represents what a multi-partial ETD comes up with by averaging the partial pitches at any given time.  That's the operation where I was curious if there was an even snazzier way of doing it.

OK, so let me clarify. 

When I said I was using the average, I was basically taking the average frequency of each partial over the entire 3-second piano note. I did this for each individual partial from each individual sensor (3 mics and one pickup). 

Here's the actual data from the C5 note. 

After I had an average frequency reading from each sensor, I averaged the four frequencies from the four sensors together to get a single average frequency for each partial. Those are the numbers on the right. You can see that the measured frequencies are all really close to each other (standard deviation under 0.2 Hz even for the worst one). 

Then I went back to the original "partials over time" data and calculated the cents offset for each time window relative to the 4 "average" frequencies in the table above. So each partial is being measured and analyzed independent of the others, and they're all being plotted on a consistent scale so you can directly compare one graph to the others. When you see one partial go up and 3 others go down, it's because that's just what they happened to do, and not a result of how I analyzed the data. 

The point of this exercise wasn't to find a way to get a stable reading. It was to demonstrate that there are better ways to graph the data than just using whatever random frequencies happened to show up in the first time window (hammer strike to 0.5 seconds) as the standard against which you're plotting all the following dots. 

There is a better way still to plot these that I didn't do, but that I plan to at some point in the future when I find the time. It is to calculate target frequencies for the partials based on an inharmonicity model of the string and plot the deviation over time from those the targets. That's what Frank Illenberger was doing in the video he uploaded on PianoWorld. (Here's a link to that: https://www.youtube.com/watch?v=lWZXgrvFolU ) 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-30-2024 17:49
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Follow-up question on this one for Anthony: In your original video you show the results of averaging the partials.  It's been a hot minute since I took statistics but, aren't there other methods that are better at reducing the influence of statistical outliers, that might give an even more stable reading?  I realize you might have just used average in the video for simplicity's sake.

I suppose that only applies if the deviations at any given partial are independent of one another.  Looking at the graphs, I almost want to say that an noticeable dip in one of the partials almost seems to be correlated with a slight rise in the others.  But I could be overthinking it...

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Apologies for any confusion -- I understand the _main_ point of the video is to show a way of graphing the deviation of the partials from a meaningful zero point, i.e. without randomly biasing them based on a single time window.  If I got you properly just now, you explained that there were a couple steps of averaging that you used to do that in the first place.  Appreciate the additional info :)

But if I understood the video right, there's a third type of average that you illustrate with a thick line around 17:45 in the video, that represents what a multi-partial ETD comes up with by averaging the partial pitches at any given time.  That's the operation where I was curious if there was an even snazzier way of doing it.

OK, so let me clarify. 

When I said I was using the average, I was basically taking the average frequency of each partial over the entire 3-second piano note. I did this for each individual partial from each individual sensor (3 mics and one pickup). 

Here's the actual data from the C5 note. 

After I had an average frequency reading from each sensor, I averaged the four frequencies from the four sensors together to get a single average frequency for each partial. Those are the numbers on the right. You can see that the measured frequencies are all really close to each other (standard deviation under 0.2 Hz even for the worst one). 

Then I went back to the original "partials over time" data and calculated the cents offset for each time window relative to the 4 "average" frequencies in the table above. So each partial is being measured and analyzed independent of the others, and they're all being plotted on a consistent scale so you can directly compare one graph to the others. When you see one partial go up and 3 others go down, it's because that's just what they happened to do, and not a result of how I analyzed the data. 

The point of this exercise wasn't to find a way to get a stable reading. It was to demonstrate that there are better ways to graph the data than just using whatever random frequencies happened to show up in the first time window (hammer strike to 0.5 seconds) as the standard against which you're plotting all the following dots. 

There is a better way still to plot these that I didn't do, but that I plan to at some point in the future when I find the time. It is to calculate target frequencies for the partials based on an inharmonicity model of the string and plot the deviation over time from those the targets. That's what Frank Illenberger was doing in the video he uploaded on PianoWorld. (Here's a link to that: https://www.youtube.com/watch?v=lWZXgrvFolU ) 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-30-2024 17:49
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Follow-up question on this one for Anthony: In your original video you show the results of averaging the partials.  It's been a hot minute since I took statistics but, aren't there other methods that are better at reducing the influence of statistical outliers, that might give an even more stable reading?  I realize you might have just used average in the video for simplicity's sake.

I suppose that only applies if the deviations at any given partial are independent of one another.  Looking at the graphs, I almost want to say that an noticeable dip in one of the partials almost seems to be correlated with a slight rise in the others.  But I could be overthinking it...

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Gotcha, thanks.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-30-2024 19:49
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Original Message:
Sent: 6/30/2024 7:28:00 PM
From: Nathan Monteleone
Subject: RE: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Apologies for any confusion -- I understand the _main_ point of the video is to show a way of graphing the deviation of the partials from a meaningful zero point, i.e. without randomly biasing them based on a single time window.  If I got you properly just now, you explained that there were a couple steps of averaging that you used to do that in the first place.  Appreciate the additional info :)

But if I understood the video right, there's a third type of average that you illustrate with a thick line around 17:45 in the video, that represents what a multi-partial ETD comes up with by averaging the partial pitches at any given time.  That's the operation where I was curious if there was an even snazzier way of doing it.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-30-2024 18:35
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

OK, so let me clarify. 

When I said I was using the average, I was basically taking the average frequency of each partial over the entire 3-second piano note. I did this for each individual partial from each individual sensor (3 mics and one pickup). 

Here's the actual data from the C5 note. 

After I had an average frequency reading from each sensor, I averaged the four frequencies from the four sensors together to get a single average frequency for each partial. Those are the numbers on the right. You can see that the measured frequencies are all really close to each other (standard deviation under 0.2 Hz even for the worst one). 

Then I went back to the original "partials over time" data and calculated the cents offset for each time window relative to the 4 "average" frequencies in the table above. So each partial is being measured and analyzed independent of the others, and they're all being plotted on a consistent scale so you can directly compare one graph to the others. When you see one partial go up and 3 others go down, it's because that's just what they happened to do, and not a result of how I analyzed the data. 

The point of this exercise wasn't to find a way to get a stable reading. It was to demonstrate that there are better ways to graph the data than just using whatever random frequencies happened to show up in the first time window (hammer strike to 0.5 seconds) as the standard against which you're plotting all the following dots. 

There is a better way still to plot these that I didn't do, but that I plan to at some point in the future when I find the time. It is to calculate target frequencies for the partials based on an inharmonicity model of the string and plot the deviation over time from those the targets. That's what Frank Illenberger was doing in the video he uploaded on PianoWorld. (Here's a link to that: https://www.youtube.com/watch?v=lWZXgrvFolU ) 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-30-2024 17:49
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Follow-up question on this one for Anthony: In your original video you show the results of averaging the partials.  It's been a hot minute since I took statistics but, aren't there other methods that are better at reducing the influence of statistical outliers, that might give an even more stable reading?  I realize you might have just used average in the video for simplicity's sake.

I suppose that only applies if the deviations at any given partial are independent of one another.  Looking at the graphs, I almost want to say that an noticeable dip in one of the partials almost seems to be correlated with a slight rise in the others.  But I could be overthinking it...

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 16:08
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

A few replies: 

"At about the 18:30 mark in your video you say "So by averaging together multiple partials, we're basically getting a stable reproducible pitch, no matter where you are in position, and no matter where you are in time."  I think your 4 charts also show that you're getting an accurate estimation of the pitch regardless of the chosen sensor as well.  Yes/No?"

Correct, for this note on this piano. I wouldn't want to make any claims broader than that based on so little data. I have sampled single string notes for all 88 keys with the 4 sensors on a couple of pianos, but it would take me a while to write a script to efficiently digest that much data, so right now it's limited to single notes. 

"Why was A2 chosen rather than C5?"

Because A2 has convenient "friendly" and familiar frequencies for its partials (110, 220, 330, 440,...) and I wanted to make the video easy to understand. 

Here are the same 4 graphs for the note C5. (I unintentionally shuffled the order when creating the collage, so while the 1st graph is still from the mic in the traditional tuning location, the other ones aren't in the same order as in my video.)  Scale on the y axis is increased by 50%.

"I appreciate that you are probably placed in a tough spot here as an ETD developer yourself, and may yet face accusations of simply promoting your own product." 

I hesitated to mention my app in the video for this reason. Also, I should be clear that I in no way meant to throw shade on single-partial ETDs. You can see in the graphs for A2 that Partials 3 & 4 are stable in position and time, and I suspect that TuneLab and CyberTuner would be using one of those partials for A2. I remember back in February you (Nathan Monteleone) did a video of TuneLab where you placed your iPad in a bunch of different locations while playing the note C4. You showed that it gave basically the same reading no matter where the iPad was located. I actually followed your lead and made a video in some client's home of PianoMeter doing the same thing, but I never shared it anywhere. 

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com

Original Message:
Sent: 06-27-2024 14:43
From: Nathan Monteleone
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Hi Peter, thanks for weighing in!  I recall similar claims were made regarding notes in other registers of the piano, so I think A2 is at least relevant.  But you have a good point that we should always take into account that results can change with any experiment depending on which area of the instrument we're working in.

------------------------------
Nathan Monteleone RPT
Fort Worth TX
(817) 675-9494
nbmont@gmail.com

Original Message:
Sent: 06-27-2024 14:26
From: Peter Grey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Why was A2 chosen rather than C5?

Peter Grey Piano Doctor 

------------------------------
Peter Grey
Stratham NH
(603) 686-2395
pianodoctor57@gmail.com

Original Message:
Sent: 06-27-2024 12:29
From: Paul McCloud
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

It's interesting that you have agreed with Steve that the ideal window of time to tune is around a half second.  The impact sound, ie., noise, is of course to be ignored.  This window of time is easily defined using another app which has a freeze indicator.  Having used this feature and tuning to this time window, I am able to get consistent and accurate results from string to string of a unison so that the result is clean without having to tax my ears.  There are some anomalies, of course, where I must use my ears.  But when I'm done, I have very few notes that require touching up.  The better quality pianos are easier, and have less false beats of course, so I rely upon the app.  Lesser quality pianos need the ear, even if tuning them is not done during the half-second window.  Doing this method has improved my tuning results, and my clients like it also.  The sound is more bell-like and "crisp" because the unison strings are tuned during the same time window.  For me, this is the key to fine tuning. 
The point of Steve's video is to show that there is a drifting of the partials at different rates over time.  Whether you tune to a later time window or not, attack vs decay, introduces a more uncertain result.  Indeed, as you pointed out, you are going to get a different measurement depending upon when and how long the measurement is done.  Most multi-partial apps ignore the first .1 second of impact noise, so that's not an issue.  But when you're tuning 3 strings individually, there is going to be a unique variation in the pitch of each of them over the time window.  Therefore, the only logical conclusion is that they must all be tuned during a period of time when the partials are most closely aligned.
There is another factor that is not being defined here.  That is the force of impact of the hammer, ie., the velocity of the hammer.  I suggest that this impact strength must be consistent as one tunes each string.  In order for me to accurately use the freeze, I will use stronger blows to get close to the desired target, but very soft blows for more accuracy.  The "fm effect" or "stronger blows create an instantaneous rise in frequency" is working here.  During a half-second interval, a hard blow is going to show up as a higher frequency than a soft blow.  I envision a "pounding device" which would create an exact blow force, adjustable of course, to eliminate this inconsistency.  Any takers?
All in all, I much appreciate the tone of this discussion.  This video does raise some questions and helps round out the topic.  Thanks very much Anthony.

------------------------------
Paul McCloud, RPT
Accutone Piano Service
www.AccutonePianoService.com
pavadasa@gmail.com

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

I found the video interesting and this offers a probable explanation of why I would find some notes on some pianos using Verituner(multi-partial) that would show an unstable display. I always assumed it was poor measuring on the displayed note or referenced notes for stretch, but unstable single partials makes sense by skewing the averaging of the strongest sounding partials over time.

In fact that presents a case against using the sensor for multi-partial tuning since Steven confirmed that the partial strengths measured at the string level don't match the partial strengths filtered through the soundboard into the room, possibly causing a change in how an app drives the display.

My takeaway from all of the graphs is to simply use a single partial with an app that measures inharmonicity of all the needed notes(to avoid the estimating that just measuring a few notes requires) and just cycle through the partials to choose the one that is most stable through the sustain of the note.

I tried that today using PiaTune on a Kimball console and ended up with the sweetest tuning ever on that little beast.

Ron Koval

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------

Anthony, I agree with others' accolades – very professional job in clarifying and correcting some of the content of Steven N's post. Like explaining the pseudo-issue of using a microphone at a certain position versus using a magnetic pickup, or the perceived need to use an ETD that does its analysis near to the attack time.

Steven, you might want to re-think your discussion of mis-tuned unisons, such as the unison you demonstrate at around 12:30 in https://www.youtube.com/watch?v=MraLJ4wx-Sk IMHO that "no" PTG tuner is going to set a unison at C5 beating at 1 bps / Hz (beats per second / Hertz) like you demo. Your unison (at 10:30) runs around 0.125 bps at C5 which is more realistic in my experience.

Also for Steven, you said earlier in this thread that in your measurements you use "fully coupled unmuted unisons," and not individual strings, and you repeat that at 2:00 in your video https://www.youtube.com/watch?v=KIxAY4u__gQ . I find the C6 (at 13:00 in the same video) to be strident and the C7 (at 14:30) to be thin, not what I expect from a 10-foot Fazioli. Might the recording be of single strings after all, or is it my hearing aids!

Finally, can someone point me to a video showing how a magnetic sensor (like what Steven sells) is used in the tuning of a smaller-sized upright? I am particularly interested in its use for the upper half of the scale.

All – enjoy Reno. Regards, Norman

------------------------------
Norman Brickman
Potomac Piano Service
Potomac, Maryland
potomacpiano@verizon.net
https://potomacpiano.com
(301) 983.9321

Original Message:
Sent: 06-28-2024 16:49
From: Ron Koval
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

I found the video interesting and this offers a probable explanation of why I would find some notes on some pianos using Verituner(multi-partial) that would show an unstable display. I always assumed it was poor measuring on the displayed note or referenced notes for stretch, but unstable single partials makes sense by skewing the averaging of the strongest sounding partials over time.

In fact that presents a case against using the sensor for multi-partial tuning since Steven confirmed that the partial strengths measured at the string level don't match the partial strengths filtered through the soundboard into the room, possibly causing a change in how an app drives the display.

My takeaway from all of the graphs is to simply use a single partial with an app that measures inharmonicity of all the needed notes(to avoid the estimating that just measuring a few notes requires) and just cycle through the partials to choose the one that is most stable through the sustain of the note.

I tried that today using PiaTune on a Kimball console and ended up with the sweetest tuning ever on that little beast.

Ron Koval

------------------------------
Ron Koval
CHICAGO IL

Original Message:
Sent: 06-27-2024 05:33
From: Anthony Willey
Subject: A review of Steven Norsworthy's YouTube video: The Piano Sound is a Moving Target

Over the last several months I've watched with interest many of the thought provoking videos published by Steven Norsworthy. Many of those videos contain bold assertions that I think could use some peer review. Since the preferred medium for Steven's publications here seems to be YouTube presentations advertised on piano forums, I will follow that format with my own YouTube presentation. My hope is that it will help people better understand some of the more technical aspects of Steven's signal processing analysis and push back on some of what I see as shortcomings in some of the techniques employed in his analyses. 

I hope you find the video informative, or at least mildly interesting. 

https://youtu.be/HNPRTHCSt2Q?si=zs3iF4wILEFbrNrI

------------------------------
Anthony Willey, RPT
http://willeypianotuning.com
http://pianometer.com
------------------------------