CAUT

Tim, 

You are referring to the latency in the display of a tuning app. I understand that. Here is the solution. The Freeze Window. It allows the user to exactly set the time window length from the hammer strike to the stop point and that is the window length and then the result of the calculations are displayed. 

In my latest video, How Spatial Position Affects Piano Tuning, link below,

https://www.youtube.com/watch?v=26zskT_GvkA&t=6s

I show exactly how a spectrogram is made. You will not need to worry about any latency issues if the app uses and displays the Freeze Window, which is essentially the tuning result from first FFT block calculation. In this example it is from the point in time near the hammer strike to the end of the block time thereafter. The samples of sound that are captured during that block of time are used in the calculation. You then see the tuning result from that block. How far off it is from zero is then used to re-strike the hammer until iteratively the indicator shows nearly zero cents while of course the continuous indicator in the app keeps moving. The jittery spreading of the partials can be seen in the continuous indicator of an app, but most apps don't show us the details that I propose in the video. Jitter and spreading is 'real physics' going in in the piano sound, and of course the ear and brain hear it too, but can't measure it with the kind of precision that a machine can. 

Steve N.