Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Novel Piano Tuning App

Gerry, thank you for the feedback. This issue seems very strange. I have not experienced this ever with the software. Perhaps we can analyze where it is coming from.
You reported these two problems:

a) Wrong note detection with a constant offset
b) Strange inharmonicity readings

First lets deal with problem a):
There are two factors which affect the note detection in pianoscope:

1. The concert pitch you chose when you setup the instrument.
I suppose you did not offset this accidentially by >50 cents. But you could check this under the ... menu > Instrument > Concert Pitch

2. The basic analog / digital accuracy of the device's audio hardware.
This is generally very good and the worst I ever experienced was an offset of 0.7 cents. You could run pianoscope's calibration against an external reference tone to check this. You get the calibration via ... menu > Settings > Calibrate

As Verituner seems to work on the device, I find it unlikely that the device has some kind of hardware defect. But to rule things out: Do you have a different iOS device on which you could try to run pianoscope as a comparison?

Problem b) is even stranger:
Even if your device had a kind of absolute frequency calibration issue, this should not affect IH measurement as this is derived from the relation of partials not their absolute values. I have no idea where this is coming from.

It might sound like a placebo, but have you tried rebooting your iPad? Here you can see how it works in different iPad models: https://support.apple.com/en-us/HT210631

Thanks for the response Frank.
Pianoscope was set to A-440. Pianoscope did give me IH measurements, but they were all a semi-tone low. Tried rebooting, even tried reinstalling software.
I'll double check settings and try another piano this afternoon. If problem persists I'll try a different IOS device.

Ah, ok. So problems a) + b) seems to be just a single problem with the absolute frequency precision. When you first started using Veritune on this iPad: Did you have to calibrate it against an external source?

Didn't need to calibrate, although I checked just to be certain.

Just a thought, Frank. Could it be a reaction to an incomplete set of inharmonicity readings affecting the tuning curve so that notes are detected a semitone low? Does your software update the tuning curve (and therefore the tuning targets affecting auto note switching) after each IH measurement? Even if it does, I assume that A4 is always anchored, so I wonder if Gerry's "and so on up the scale" refers to behavior all the way to A4?

As Gerry's problem with wrong note detection already occured as he was measuring the IH, the tuning curve calculation couldn't have been the culprit as it only happens AFTER the IH has been measured. So either the piano was more than 50 cents detuned (what he and Veritune most certainly would have noticed) or there must be some other factor which we do not understand yet.

I tried Pianoscope on my iPhone this afternoon and it worked perfectly. It's probably safe to assume that the problem is in my iPad. Although it just seems odd that Verituner seems to be working O.K.

Thank you Frank, for this full-featured beta test. The app is well-designed and documented, and I find it intuitive to use.

However, I can't get Pianoscope to calibrate my iPad to a known 440 Hz reference, it shows up as too flat. The adjustment maxes out at only 2 cents, should it be 100 times that for 2 semitones? Pianoscope detects that 440 Hz reference as 403 Hz, or -152 cents.

Tried powering off the device, restarting Pianoscope, taking off the case. Nothing helped. It's an iPad Air 3rd generation.

For comparison, I tried Anthony's app PianoMeter on the same device. Without calibration, it detects that 440 Hz reference as A4 (A440), -2.3 cents flat.

Gerry & pianofish: I don't think your devices are at fault. There must be a bug somewhere in my code that only occurs with some devices. Perhaps the audio system delivers different sample rates than I expect. I need to look into that.

Gerry: What model is your iPad?

Question:

Why are the IH measurements so quick? Almost instantaneous, in fact. IH measuring seems to be a much more lengthy process in other tuning programs, particularly below c4, as you probably know.

Hi Frank -
iPad mini - 5th generation. Running version 14.4.1

I suspect this may be the issue. I think there are some iPads that only support a 48 kHz sample rate. Not sure if you're trying to sample at 44.1kHz, but the ratio between 48 kHz and 44.1 kHz is close to the ratio between 440 Hz and the 403 Hz number pianofish reported.

As it turned out, in the setup of the audio stack, I used the sample rate of the audio OUTPUT stream to configure my parameters instead of the INPUT stream. On most iOS devices, they seem to be identical so that this error did not have any consequences. On Gerry's and pianofish's iPads however, the output stream seems to default to 44100 Hz and the input stream to 48000 Hz.

Gerry & pianofish: Please check out the new build 414 I have just uploaded to Test Flight. I think the bug you encountered is fixed in this release. But please test it with your iPads as I do not have any device with which pianoscope showed this problem.

Today I was a bit playful and I tried to implement an experimental feature of which I don't know if I should put it into the final release. I don't know if it is of much use to experienced tuners but for students it might be very helpful. I am curious what you think. The feature is currently not documented in the manual.

Replaying recordings of selected partials:

If you enable the partial spectrum via the settings, you can now tap on the word "Partials" above the keyboard in the tuning view. This opens a menu which allows to switch between two modes of operation. The default mode is the "Continuous" mode, which simply shows the partial intensities live as before. The new mode is currently dubbed "Maxima per Note". This freezes the partial intensities at their relative maxima until you play the next note. When you now tap on one of the frozen partial bars, pianoscope will replay a recording of the last note, but band-pass filtered for the partial you tapped on. This feature may help novice tuners in learning to hear partials and to isolate beats. It works best with headphones as the speakers of iPhones and iPads are not the best.

That was fast Frank! Build 414 fixes the issue on my iPad. Now the A440 is measured with better than 1 cent accuracy before calibration.

Thank you Frank. I do not have any tunings scheduled tomorrow, but will try it out Friday.

Thanks, I'm glad to hear that.

Let me correct/refine this statement. Some tuning programs are FAST at arriving at an IH, others are slow. Is it the case, perhaps, that experience has shown that there is effectively no real advantage to doing multiple IH measurements? Or, perhaps, doing multiple IH measurements and averaging them? Perhaps one IH measurement is all that is really needed (unless, of course, it appears to be way out of range, in which case another is required).

When designing a system for measuring frequencies you need to make many tradeoffs. If you use FFT, you need to choose a sample rate, a buffer size, a windowing function, an interpolation scheme and so on. And they all influence latency and accuracy. As the frequency range of the piano is quite large, there is no single perfect set of parameters. pianoscope is adaptive and very good at reducing latency, as it does not only rely on a streaming buffer which is analyzed ~5 times a second but instead is event based and can react very fast to audible changes.

But the question also is: How much information do you need about a note in general? The inharmonicity of a piano tone changes with its envelope duration. Which inharmonicity do you want to use for tuning? Which one do you use when you tune aurally? Do you use the early, middle or late phase? If you use the early phase, you give more weight to the lower partials in determining the inharmonicity as they are fading out faster than the higher ones. From my experience, if you consistently take just the first 1.5 seconds, you are getting good tuning results. But I think you have to be consistent. Pianometer for example, when the tuning is not yet locked, listens and measures all the time. So it highly depends on you how long you play every note.