TestFlight Beta of pianoscope 3.0

Abraham Lee

Great work, Frank! I'm unfortunately stuck on an older version of iOS on my phone, but will definitely upgrade when I can.

Frank Illenberger

Abraham Lee Like pianoscope 2, version 3 is running on iOS 15 or later. It is only the Time Charts feature which requires iOS 17 or later.

MilanDka

Dear Frank.
I have a few translate mistakes: Prefer built-in mic “Preferovat vestavěný mikrofon” and Use multiple mic “Použít více mikrofonů”, false beats “falešných záznějů”. The last one is part of the description of the function of using multiple microphones.

Please replace the word ‘Inharmonie’ with the word ‘Inharmonicita’ in all parts of Pianoscope. Because: in simple terms, inharmony is the problem, while inharmonicity is the way we measure or describe this problem. It is Czech language. For us, it is sometimes difficult too.

Are you sure that the iPhone 11 doesn’t have multiple microphones? I can definitely see three: the main one near the Lightning connector, the second one near the camera lenses, and the third one near the top speaker. The “Use multiple mic” option is grayed out for me. Or is the hardware so weak that it can’t handle using multiple microphones?

Frank Illenberger

MilanDka Thanks. I will correct the translation mistakes with the next update.

Some older phones have multiple mics but the operating system nevertheless does not allow apps to access them individually. So if pianoscope does not offer the option on your iPhone 11, we are out of luck.

Phil

the search functionality does not work at all for me on my iPhone13mini.

I can't find any tuning file by manufacturer, serial number, etc. neither in the Files app nor in spotlight search nor inside the pianosope app.

I have set up Pianoscope to save all the tuning files in a Nextcloud folder that is synced to my Macbook. That's all working fine and on my laptop I can find tuning files by serial number in the Finder App, but the same search does not find anything on my iPhone.

Any idea?

Frank Illenberger

Phil The files need to be indexed by the Spotlight daemon on the system. Sometimes I observed that the system simply does not do that. One thing you could try is to duplicate the folder with your tuning documents in the Files app under iOS. This could trigger the system to reindex the files.

JBrasell

Frank Illenberger Same issue on iPhone14. Tried duplicating the pianoscope folder in files app. Still not able to search by other fields.
Will it possibly work with new files created?

Burkhard

Hello Frank, V3 is doing great so far.

I like the new algorithm very much.

The measurement for a new high voice seems to be a bit sensitive and needs a little more rest and time than before. Can this be related to the number of microphones?

Will Pianoscope 3.0 automatically convert already tuned pianos to the new algorithm?

I now have several hundred data sets and don't want to deal with it for long.

Frank Illenberger

Burkhard Currently, you have to manually switch it to v3. But you are right, there needs to be a quicker way to do it. I will think off something.

Could you do a screen recording with sound for me of the sensitivity problem syou were encountering?

Frank Illenberger

Burkhard In the latest build 1594, pianoscope shows an alert when you open a tuning document that still uses the v1 or v2 tuning curve algorithms with the factory tuning styles. In the alert, you can upgrade the document with a single tap to v3. The alert is only shown once for every document, so if you decide to stay with an older algorithm, the app does not nag you again.

Burkhard

Moin Frank, what is the function of the "analyzer" in the toolbar?

Frank Illenberger

Burkhard This is an internal debugging tool which should not appear in the toolbar. I will fix that.

MilanDka

Hi Frank,

I was sick with the flu for a week and had plenty of time to think and study what the best tuners actually listen for when tuning. I came to the conclusion that the most important thing is to listen to major thirds and sixths. Then I also discovered certain patterns in tuning this way, including the expression of interference beats in these intervals.

When tuning a piano using only this method (I’m no longer capable of doing it myself, but my colleague is truly an outstanding tuner), the fifths, fourths, octaves, and tenths come out exactly as they should. However, there is a significant difference in the thirds and sixths compared to all tuning software. The only one that supposedly comes somewhat close is Reyburn Cybertuner. This is not just my own idea—my colleague and I discuss this issue a lot.

The problem with all tuners is that they make the thirds and sixths too fast, which results in a very harsh tuning. I’m attaching a table of interference beats as well.

With Pianoscope, I still have an issue—even in the Beta version—starting from G#3 downward (in most cases, Pianoscope places these tones too high; when I tune this area by ear using major 3rds and 6ths, I place the notes a bit lower) and from A#4 upward. Going towards the bass, the lower register ends up lower overall compared to Pianoscope, and the same applies to the treble.

Do you think it would be very problematic to create a new tuning style based on this principle? I don’t know to what extent Pianoscope accounts for this, but it seems to me that it doesn’t much. Maybe it’s just my impression. In temperament, we are fairly close, but in the areas I mentioned, not at all. The result of this tuning is that the whole instrument sounds more harmonious and pure.

I mean that Pianoscope would first calculate the exact frequencies of each note starting from a reference pitch, such as A4, based on precise interference beat rates. Then, for each note, it would display whether the number of beats is too low or too high—just like it currently shows whether you’re flat or sharp.

Explanation:
Number of Beats is determined by the frequency difference between two tones.
This difference (in Hz) is perceived as the interference beats, which are changes in amplitude that the human ear detects as a rhythm or pulsation.
Number of Beats therefore represents how fast the tones are oscillating in a way that creates a perceptible "beat" to the ear.

Then perhaps it wouldn’t even be necessary to record each instrument to determine inharmonicity—it’s just a hypothesis, because all the intervals would essentially "fall into place" on their own.

Don’t take this the wrong way—you’re doing a great job, and I’m very happy with Pianoscope. I’m just looking for ways to improve it.

Rick

MilanDka I might be wrong but your tables don’t factor in inharmonicity.
Rick

Frank Illenberger

MilanDka I cannot find anything wrong with the beat rates of thirds, sixths and tenths in pianoscope tunings. Here for example, you have their beat rates charted for a small upright piano:

To my mind, everything is as it should be:

The beat rates of thirds, sixths and tenths are smoothly and progressively getting faster.
Sixths are beating slightly faster than thirds, which is equivalent to wide fourths and narrow fifths.
In most of the range, thirds and tenths have the same beat rate, which is equivalent to beatless 4:2 octaves. Only in the low bass, their beat rates need to be different, because you don't want beatless 4:2 octaves here in favor of less beating 6:3 and 8:4 octaves.
The third F3-A3 beats with 6.9Hz which is very close to your desired number of 7Hz, although I doubt that absolute magic beat values are important.

In the following image you see the same graph for a large grand piano. Here, the beat rates of thirds and tenths can be the same in the low bass because you can make the 4:2 octaves beatless because of the low inharmonicity of the long bass strings of this instrument which results in the beat rates of 4:2, 6:3 and 8:4 octaves being nearly identical in the low bass.

Pianoscope does not construct tuning curves like an aural tuner, and it does not have to. It does not "listen" to fast beating intervals, but instead always focuses on the slow beating intervals. Mathematically, this is equivalent as the beat rates of only slow intervals fully define the ones of fast intervals and vice versa. For the human hearing, it is easier to listen to fast intervals, but for a machine it does not make a difference.

MilanDka

I don’t think so, Rick. Inharmonicity simply cancels out if you strictly follow the beat rates. Of course, the software must account for how the temperament is tuned—whether it’s stretched more or less—and adjust accordingly in both the bass and treble.

The number of beats is simply independent of inharmonicity. For example, the F3-A3 third must always have around 7 beats. If it doesn’t, it sounds disturbing to the human ear.

Another thing I’ve noticed: On very small instruments, from E1 downward, it seems that in the Beta version, Pianoscope starts listening to completely different intervals for tuning. In the bass, it’s crucial to listen to 10ths and also 12ths. If those are hard to hear, 17ths can be used instead.

MilanDka

Dear Frank,

Thank you for the explanation and the clear demonstration. I just have a few questions:
1. How did you determine from the graph that F3-A3 beats at 6.9 Hz?
2. Why do I slightly diverge by ear in the A2-F3 range? Here, we tend to go 0.2-0.4 cents lower by ear compared to Pianoscope.

I can only explain this by assuming that the ear perceives multiple influences—various partials or possibly false strings—which slightly adjust the final tuning and tonal color.

Regarding A0-E1 (perhaps up to G1) on very small instruments: Before I got sick, I tuned a second S144 grand and a 106 upright. Both had the same issue. The thirds and tenths were fine, but due to extreme inharmonicity, it just didn’t fit, so I relied more on 17ths, 19ths, and triple octaves. Unfortunately, I didn’t create a static tuning.

I understand that this is probably difficult to program, but in this version, you started using AI for better results. Could you take a closer look at this area?

Edit: I think I’ve found another mistake. In the tone generator settings, I don’t see the option to select Spectrum – Measured Partials and Pure Sine.

Thank you.
With respect, Milan

Frank Illenberger

MilanDka

If you have only activated a single tuning interval in the deviation curve view, pianoscope will show the numeric beat rate:

From afar I can't tell why your ears differ from pianoscope's deviation display. Surely, I cannot rule out that there is some kind of bug in pianoscope. What you could do in such a case is to send me the tuning document and audio recordings of the individual notes in the interval. Then I could do an analysis.

If you have done a tuning with pianoscope in which you don't like certain intervals, simply send me the tuning file and tell we which intervals are problematic for you. Then I can try to understand what you expect and what pianoscope is doing, I only need a recorded static tuning if you do aural corrections.

I have removed the tone generator option for Pure Sine. You can achieve the same effect now by setting the max partials count to one.

Frank Illenberger

MilanDka I had another idea. The reason for the problem you are encountering might not lie in the tuning curve calculated by pianoscope, because it looks very consistent. Instead it may lie in the inharmonicity modeling. Can it be, that the piano you were tuning had bass strings of poorer quality? As an experiment, you could try to increase the inharmonicity weighting to 100%. There is a small chance that it could improve things. All further improvements to inharmonicity modeling must wait for version 4. I already have some ideas for it.

Burkhard

Frank Illenberger Perfekt 🤩

Burkhard

Moin Frank, the use of several microphones slows down the measurement of the inharmonicity and pitch on my iPhone 15proMax.
If I switch to one microphone, it's smoother. Would it be possible to use only one microphone and to vote everyone for the tuning?

Frank Illenberger

Burkhard Thank you Burkhard. This seems to be a problem with background noise. I will look into it.