Pianotech

Expand all | Collapse all

Documented Mics and Placements Affecting the Frequency Measurements

  • 1.  Documented Mics and Placements Affecting the Frequency Measurements

    Posted 01-12-2024 21:11
    This is documenting the fact that, indeed, with two parallel recordings of B1, post processing them in Matlab to extremely high resolution frequency analysis, harmonic by harmonic, the results are rather startling. The sensor vs. the mic are not picking up the exact same frequency. It also changes substantially by mic position. The sensor is invariant. Not to question the interpolation in my Matlab analysis, the frequency interpolation is good down to a few parts per million, micro cents that is. You can clearly see that there will be an ETD recorded difference (depending on how good the ETD's frequency measurement algorithm is), of more than 1 cent on some harmonics. The standard deviation of the first 6 harmonics is 0.65 cents in this example.
    Prof. Norsworthy 
    Mic vs Sensor Frequency Differences
    B1 Sensor vs Mic Spectra



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



  • 2.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Posted 01-12-2024 22:27

    The current state of the art results in a bloody subjective mess of opinions and variances and preferences, unless we use an invariant data-driven analysis.

    Even if we count beats for the intervals, we still don't know down to 0.4 cents by ear what we are hearing! Plus, the frequency is drifting while we are beat counting if we listen for more than 1 sec. This is a fact. It seems to be ignored in the community. Even if we use the best freq. analysis in the ETD, we cannot trust the frequency readings due to mic variance. So I  decided to take a different approach and do super fine post analysis of the freq difference of the mic and sensor from the same attack and record them. This will tell us what we need to know and hopefully create a gold standard of invariant measurement for a new approch.



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



  • 3.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Registered Piano Technician
    Posted 01-13-2024 00:00

    The current standard is not counting beats, it is checking for a smooth progression of beat rates as one moves up or down the scale using various intervals, this guarantees that you have tuned an ET scale. This is an aural check that even ETD users use as a final determination. Music is an aural medium so ultimately the tuning must satisfy an aural test. If robots start showing up to concerts, perhaps we'll have to reconsider.

    It seems the method you are prescribing exceeds the ability of either microphones or the human ear to discern. 

    If you want that level of discretion, why not just use a computer modeled digital piano sample.

    Pianos breathe, under most circumstances, in most environments, pianos will drift outside of your parameters within hours if not minutes.

    One degree of ambient room temperature can change a long wire in a large piano over 1/2 cent. These short term changes do not affect all strings evenly.



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



  • 4.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Posted 01-13-2024 00:07
    I am offering a major reduction in variance, plain and simple. Repeatable, invariant. Why start with an error? Room temperature and humidity, I get that, but I am offering a way of STARTING with less error.

    Make sense?




  • 5.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Registered Piano Technician
    Posted 01-13-2024 00:21

    I understand, but I suspect your precision exceeds the tolerance of both the system (piano) and the listener. 

    And there are so many other variables from the instrument, to the style of music being played, to the room acoustics, etc.

    Musical instrument design itself is a long series of compromises and then one adds the environmental variables. The piano technician's practice revolves around negotiating these compromises. Choices, often subjective, must be made. I.e. a stronger attack comes at the expense of a shorter decay, a longer decay comes at the expense of a weaker attack. 



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



  • 6.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Registered Piano Technician
    Posted 01-13-2024 00:25

    By the way, as was mentioned by someone in an earlier thread. It is not settled fact that a perfect unison is optimal; not among tuners or concert artists. I can attest the latter myself. And this can change from piano to piano.



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



  • 7.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Posted 01-13-2024 00:48

    Why in the heck do you desire to introduce error from the outset? Really?



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



  • 8.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Registered Piano Technician
    Posted 01-13-2024 01:08

    The observation is that sometimes the tone disappears after the attack with a perfect unison.

    Some people also talk about a "bloom" effect with unisons tuned a certain way where the tone swells after the attack.

    Perhaps this is a conflict between engineering values and musical values. Just be glad we aren't talking about accordions!

     



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



  • 9.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Posted 01-13-2024 01:19
    The bloom is deceptive. It is a half wave crossing at the primary beat fundamental, say the beat is 1 Hz, then the first half wave is 500 ms, and then the bloom is 1 sec, etc. I have a model for this phenomenon. The ‘bloom’ is deceptive and psycho=acoustic and we only think it is a bloom when it is just a realignment of the next 180 degree phase of the best. It’s bogus. I play this for my class in my lecture. I show the waves and play the audio files of the mis-tuned unisons. I show them the math and they hear it as the graph shows.




  • 10.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Posted 01-13-2024 01:22
    False Bloom


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



  • 11.  RE: Documented Mics and Placements Affecting the Frequency Measurements

    Registered Piano Technician
    Posted 01-13-2024 00:27

    Although this looks very technical, and it is, there is a profound meaning and truth here.  If you examine the two graphs, the upper one with the sensor and the lower one with the mic, you can see clearly that the upper graphic shows much cleaner spikes, and that they are equally spaced, and they are taller.  What we are seeing  is that the mic picks up the information from the sound pressure through the air, and the sensor only detects only the movement of the string.  The difference between these two sources is not just loudness, but frequency also.  This is because the mic can't pick up and distinguish the frequencies like the sensor can.  They are "smeared" by the mic, and so the ETD can't measure and indicate the exact frequency. Notice in the lower graph that there is a spike between the first and second partial that isn't there in the image above.  The first partial is the first spike from the left-to-right, and there's no box for it.  The ETD is going to try to "use" that spike in the calculation of the target frequency.  Oops!

    The information that is most important to know here is in the boxes showing X and Y axes.  Click on the images above to clearly view the information.  The "X" number is frequency, the "Y" is amplitude.  Both amplitude and frequency are different of the mic vs the sensor.  But the fact that the frequency is different is the most important take-away here.  If you look at the table, you can see differences as much as 1.49 cents on the fundamental, and other differences of the higher partials.  This is why using a sensor is much superior to using the built-in mics or even the best external mics. 

    Unfortunately, this is not intuitive to our minds.  A mic is very good, right?  We make nice recordings with them, and they sound amazing.  But for creating an accurate piano tuning with an app, the mic is not ideal.  The soundboard and bridge arrangement is like a filter, where the board is more efficient on certain notes and makes them sound louder, and others softer.  The acoustic signature of the piano structure is not smooth, which is one reason we do voicing.  We need to try to eliminate this inequality of sound pressure by softening or hardening the hammers, and other procedures.  But even the best pianos are going to have volume and tone variations note to note, section to section.  All of this is affecting the sound coming from the piano, but  is undetected by the sensor.  Yes, it will detect a change in amplitude of a certain partial if you voice a hammer, but otherwise the sensor is "deaf" to the sound of the piano. 

    For example, when you make a recording, you want the raw track to contain all the information.  Then, using plug-ins, or outboard gear, the original waveforms are adjusted and edited.  But if you use the plug-ins and other gear in the original track, you can't remove it.  Similarly, we don't want the filter of the piano soundboard to affect the real information coming into the ETD.  The only way to get the best input to your ETD is through a sensor picking up string vibration.

    One question may come up, which is, "Can't I just get a guitar pickup and make it work?  Why do I need the PIanosens?"  Over months of testing, we met problem after problem doing just that.  These guitar pickups are prone to noise, hum, and the magnetic pull on the strings.  It is well known among guitarists that the pickup you use will affect the tone.  Pickup makers have dozens and even hundreds of designs, which cater to the taste of the player, as their output is not equal over the frequency spectrum.  In the end, Steve collaborated with an engineering firm who specializes in magnetic sensors in the aerospace industry to produce a product that has an even, uncolored spectrum, and is not affecting the string with magnetic pulling.  They use million-dollar machines to wind coils with wire that is hair-thin, and make thousands of turns of that wire on the core. There is no hum or noise being picked up by the sensor, or at least the noise floor is almost not detectable. It is crucial that the noise floor is very low, so that the ETD can better detect the string vibration and not be affected. 

    Please also note that the graphs are not produced by any of our ETD's, but rather in a program called MatLab which is widely used by engineers, professors, and other companies around the world which have capabilities beyond what our cell phones and ipads can do.  It is a laboratory standard in these fields.  Steve has spent countless hours making the videos and spectral analyses on the subject of piano technology, making sure that any of the concepts he is presenting is impeccable and proven beyond any doubt.  No one has done this kind of analysis before, partly because it would take too much money to pay someone to do it, and the tools weren't available that could do the job.  His vast experience in the communications industry and signal processing has allowed him to make discoveries in our field that no one else has been able to make.  Sometimes his presentations are a bit too technical for those of us who aren't engineers, but I have learned a lot from him over the past year.  I hope these concepts are made a little more clear by my explanation to my fellow laymen.



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