Thanks so much for posting the link to these tunings.

I admit, given the earlier concerns in this thread, that I'm still mulling over what we can do with these charts.  Let's see...

1. We can change the string length and tuning in PT so that the IH is changed and the partials match up as per the charts. Not easy, but it can be done. Would help if we looked up the string length for each make.

2. These charts might be used to dev entirely new ptq's? Don't they serve to tell us the IH of each single string note and duochord (?) and trichord of these many, many instruments? Hm....It's only 160 pianos. Philippe, Nic, Julien, do you have a few minutes...? EDIT: Well, they would give us the iH of one partial. Not sure how much the iH of other partials could be guessed from this.

3. We can apply these tunings randomly to hear the results. Some are bound to have interesting sounds and or serve as points of departure.

4. They also could serve as raw data that would allow one to learn the range of IH on each note for different makes?

Bob, do you have other thoughts? Skip? Gilles? Philippe?

Every now and then I get a referral from this forum regarding tunings I have on my site for downloading.   I wanted to update that information.

I know this forum's 'thread' has been inactive for a while, but I want to let whoever might be interested know that I now have over 160 tunings available for downloading from my site.  These tunings are available in both the Excel format and the .SAT format.

Here is the link to the download page:  http://pianotuningtucson.com/pianotuning/tunings/   (You will need to fill out a little form before downloading, but otherwise, the  download is free.)

An accompanying  README file will be downloaded with the tunings, with some notes and suggestion as to how to use these tunings, what the information contained on the header line (title) is, and how it can be used to pick a tuning to match the piano being tuned. 

Another bit of info which might be of interest is the partials used in these tunings:
A0 - G#2 use the 6th partial
A2 - A4 use the 4th partial
A#4 - C8 use the 1st partial (fundamental)

This partial arrangement has a few significant advantages:  One of the most notable is that now there is a double octave from A2-A4 with both notes using the same partial.   The other significant change is using the fundamental from A#4 - C8.   

These tunings of course, are intended to be starting points, and were derived from measurements made on real pianos.   Since I use an AccuTuner III, these tunings are particularly suited to the SAT III or SAT IV, since these 2 models of AccuTuner have the Double Octave Beat (DOB) feature.   (This DOB feature allows the memory tuning to be stretched or contracted wherever needed to most accurately fit the tuning to the piano.)

Again, thanks for the referrals, and your interest.

I look forward to following this forum in the future.

Bob Conrad
Tucson, AZ
http://pianotuningtucson.com/

Greg.

Yes it does, and this is exactly how the piano tuning is programmed inside Pianoteq for all temperaments except the flat temperament.

Oh!    I'm puzzled about the Octave Stretching setting though, in that case.  I thought that when a real piano tuner tuners a piano, the degree of stretching was mostly defined by the inharmonicity, which is a constant.  For example, a piano tuner once told me that "ALL pianos are stretched tuned, because of the inharmonicity". How does the Octave Stretching setting relate to how a piano would "typically" be tuned? If the slider is all the way to the left, does that represent some base level of stretching that is necessary for the piano to sound good, for a given amount of inharmonicity?

Greg.

I am only very loosely following this thread, but perhaps what might be interesting is for Pianoteq to have an automatic tuning mode, where for a given piano/string length, it will be tuned as if a real piano tuner had tuned the piano by beat tuning - does that make any sense?
Greg.

Yes it does, and this is exactly how the piano tuning is programmed inside Pianoteq for all temperaments except the flat temperament.

I think it worked because the default M3 piano is a good fit--it's a modern grand, and the D and M are modern grands, so the string length is about the same. I just used the D's stretch. Pulling the fundamentals around from the flat equal temperament pulled the partials into shape because that's what the cent changes were meant to do. But I may get lost trying to work with those partials in the spreadsheet. But if the string inharmonicity stays as narrow...

I agree that a way to store beat tunings would be great. Would that be the same as storing stretch tuning templates? It's almost unethical, since a real tuner apparently measures the inharmonicity of each string to tune it and others against it. But Pianoteq lets us change the string length and tension, so it would be possible, I think, to bring up a tested stretch tuning for a specific Kawai, etc and adjust other things.

Sorry--a long post about what's probably self-evident.

Do you know anything about Scala files?

Sorry this is a bit late - only saw this question just now for the first time. But - no, I do not.

Regarding string length, even if it were possible to match them very accurately, even then I suppose there's no guarantee that the inharmonicity would exactly match, because that also depends on the string material and how their wound etc etc, AFAIK. 

I am only very loosely following this thread, but perhaps what might be interesting is for Pianoteq to have an automatic tuning mode, where for a given piano/string length, it will be tuned as if a real piano tuner had tuned the piano by beat tuning - does that make any sense? If some piano tuners use slightly different methods of tuning, perhaps Pianoteq could even incorporate these as well.  Note that I have not attempted to comprehend Joe's recent post on tuning yet.

Greg.

The partial changes occur at C3, C5, and C6.

The partials referenced are:
1. The double octave-fifth (partial 6) in the bass at A0-B2
2. Double octave (partial 4) C3-B4
3. Octave (partial 2) from C5-B5,
4. Fundamental (first partial) from C6 on up to C8.

See the manual for the electronic tuner sold on the site for more information about how how the numbers were derived. There is no other information there on tuning by partials, but there is plenty on the internet... (This chart also lists the serial number for each piano, by the way.)

But the Steinway D and M tunings were correct: those were derived from the other chart, which listed cent deviations from the fundamental.

And it looks as though Scala won't help: If I'm reading its help page correctly, you have to have regularly repeating octave relationships. You could go in and create one stretched octave, but then you'd have to manually edit every other octave in the PTeq preset. So for now all we can do is manually create a tuning and save it as a preset, as far as I can tell.

Jake,
I suspect that you and most of us probably know this, but my understanding is that you'll also have to ensure that the frequencies of the partials match the Steinway's, too, in order to fully recreate the tuning. I assume this would mostly be accomplished by setting the string lengths accurately. (?)

Greg.

Skip:

I at first tried to work with the string length, but gave up. Partly because I'm not sure of the default settings: Clearly, not all of the strings are by default the same length. The numeral for the default setting must be for the longest string. So each change we make in the NE pane appears to be relative to an unknown setting per string? Not a rant--I just wish the NE pane reflected the length of each string...However, we can of course see the pitches of each partial in the spectrum profile, so we can check the results of stretching the strings there.

Do you know anything about Scala files?

Yes if, with the C3 solo recording preset, I move note C3 up a cent and E3 down a cent (and put reverb off) and play both notes together, I hear a very slight difference in spectrum color, but not a difference in frequency...But this a 2 cents difference.

For me, the slight differences in spectrum color add up if applied across the keyboard.  But I'm probably paying too much attention to the sub-decimal settings. My thought was just that since they were available, and have a (still slighter) effect on the tone, that I should try to get the tuning exactly the same.

But this a 2 cents difference. I just found this site: http://hyperphysics.phy-astr.gsu.edu/HB...ts.html#c4
which speaks of a JND (just noticeable difference) being 5 cents on average. The frequency differences between two tones moving up the scale corresponding to this JND vary a lot of course.

I recall some years ago using this test :
http://tonometric.com/adaptivepitch/
which uses pure tones of varying frequencies (I don't recall what was the final frequency but probably in the 1-2 kHz range) where I found out I could discriminate pitch down to 1 Hz if I remember well (if the test frequency is 1 kHz that would give a JND of 1.73 cents)

Maybe you could try this test to see what your JND is. The test is longer if you have good discrimination, so I won't redo it now, but it is interesting.

By the way, don't worry about your worrying...I just wanted to save you some time if possible.

Another thought, I find it funny having to discriminate between the preset C3 and the same note...maybe as the model improves and we get beyond the (Y)C7 no more precautions will have to be taken.

I should just use other keys in the meantime...

You probably have better (or younger) ears than me because if I use the calculator site you gave, adding 1 cent everywhere for example is equivalent to tuning to 440.25 instead of 440 Hz, and I can't hear that difference, try as I may...

The fact that the diapason field displays only 2 decimals and the detune pane is limited to integer values probably means higher resolution is not very important.

Blowing up the pane indeed shows a graphical display change while the value stays within an integer cent, but I don't think any decimal cent value is calculated in the model, it could be rounded to the integer(but I might be wrong of course...)

If there is, well I just don't hear it.

But do you hear a difference if you add a cent here and lower a cent there, if you play the notes in chords? (Most often, though, the changes in a stretch tuning are larger, by as much as - or + 18 full cents in the extreme ranges. I'm probably being too worried by wanting to control to the decimals. But I do worry...)

The fact that the diapason field displays only 2 decimals and the detune pane is limited to integer values probably means higher resolution is not very important.

Blowing up the pane indeed shows a graphical display change while the value stays within an integer cent, but I don't think any decimal cent value is calculated in the model, it could be rounded to the integer(but I might be wrong of course...)

If there is, well I just don't hear it.

And each note is poised in chords with other notes that may be slightly detuned to match in a stretch tuning. And the exact decimal tuning has a big effect on the sound created by the strike point, the damper release, and the string length. A .10 movement can bring out some strange partials, from what I can hear. Please understand that I'm not trying to be doctrinaire or pretend to have specialized knowledge beyond what I'm discovering as I work. I was surprised by the strong effect of a few notches...I'm still learning all of this.