You have to consider many a historical piece was written for piano, another for cello, and still others for either violin or viola, today you obviously write for smartphone instead.  You do if you tune those instruments to a smartphone app.

A lot I’m going to say about just intonation, just later!

It is the out-of-tuneness of equal temperament, and its complex harmonic nature, that made the music of classical and romantic period composers such as Chopin, Ravel, Debussy, Tchaikovsky, and Rachmaninov possible, as well as all more modern music including jazz and any music that changes key or tonal center (sometimes as frequently as the base-note and chord is changed in music).

So we can thank out-of-tuneness for a lot, if we are inclined to see the beauty in what might be considered "imperfection."

https://en.wikipedia.org/wiki/Enharmonic_keyboard

While none of these solutions saw general interest or mass production, they are an interesting window in virtual solutions that can be created in PTQ with advanced scala files that would make for practical realizations of these--otherwise--outlandish ideas for moving away from fixed equal temperament towards total realtime control of tuning like a violinist has.

I would be curious to see in five to ten years if technology like the Steingraeber transducer "hybrid" with PTQ and similar projects solve some of these centuries-old puzzles for achieving actual realtime "perfection" in tuning in a wide variety of musical applications and situations.

I failed to see the humor you had originally intended in your twice previous post!  You are absolutely right:  if someone is going to acquire a $six-figure piano or a Strad violin, of COURSE they would not use a $4.00 phone app to tune the instrument.

Several years ago, just for fun, I purchased a Korg hardware guitar tuner and another brand of hardware tuner (each~$25-$20) that has a piezoelectric pickup, both equipped only with equal temperament tuning, which is ironic for the tuning of guitars.  Both tuners were fun to play with, but their accuracies of display were nowhere near what was required to tune a piano with any type of good sound in all keys.  I actually gave one of them to my Church's choir director, because he wanted to have a hand at tuning his own piano.  I told him that, at best, he might be able to lay the bearings (tune the middle one or two octaves) of his own personal piano, but that he would have to tune the octaves by ear.  I never heard from that person any results he had gotten from trying to tune his own piano.

It took me approximately five years (1968-1972 or -73) before I was able to tune a piano with enough assurance that the instrument was in tune, to the extent I could tune other peoples' or institutions' pianos.

Cheers,

Joe

Personally, I'm seeing right now some humor in a likely typically modern scenario in which a guy may want to tune a grand piano (valued at possibly $80K) or even someday a Stradivarius, hardly to a temperament or to other musical instruments in the hands of human beings (God forbid)  —but to a $4.00 smartphone app he just downloaded.

At some future date any temperament as it pertains to a human condition like a human frailty, possibly will have lost its original true meaning as it becomes surrounded only in a high tech age by lifeless cold steel and heartless machines.

Now I feel badly.  I've allowed myself to become defensive, about my taste and anything funny to me.  Let me examine myself.  And, let me decide if something is really best for me.

People, a guy wants to just tune maybe a Steinway, Bechstein, Ant. Petrof, Bluethner, Steingraeber, or any other grand piano, to his phone app, no big deal!

That's quite alright with me that maybe a guy just wants to tune a piano to his phone app.  Agreed, no big deal.

This thread now has over twenty responses because the original title was "Piano always out of tune".  Some people wanted to help; others wanted clarification; still others offered explanations.  The best part of this discussion is that many ideas were floated and the topic was discussed with dignity and respect for fellow Pianoteq forum members.  No flaming has occurred, nor does it need to occur.

Cheers and all the best,

Joe

I also notice "Octave Stretching" doesn't go below 0.95, a setting of zero would mean no stretching right?

Melodyne from a Dutch based Celemony does compensate for any octave stretching on tracks throughout a mix.  If you choose, it will remove it entirely or adjust other track instruments accordingly.

People, a guy wants to just tune maybe a Steinway, Bechstein, Ant. Petrof, Bluethner, Steingraeber, or any other grand piano, to his phone app, no big deal!

The way I used to work with sampled piano libraries would be to play a given piece, live, while recording a backup copy to midi, and then reviewing the midi file afterward in a DAW (usually Digital Performer Version 5 or Logic Pro Version 6 at the time).  Whenever I would find notes that sounded distinctly out-of-tune, I had to manually find the note in the DAW, and modify the note-on velocity to fire the next nearest same-note sample that was not still out of tune.

When I acquired Pianoteq version 3 in the first decade of the twenty first century, I felt terrifically relieved in NOT having to edit specific note-on velocities that happened to otherwise trip out-of-tune samples.

Short answer:
Other pianos, including sampled pianos, do NOT have a better level of tuning accuracy than Pianoteq.

Yes, that does demonstrate well the advantages of the Pianoteq engine.

Gives great context to the control for "Unison Balance" in Pianoteq.

We can alter it for the whole piano in Standard and Pro: right-click "Unison Width" to find it.

Or any strings or ranges, as one of the editable features in per note edit mode in Pro.

Altering unison balance seems to shift bias a little from the 'centre string', or theoretically the central 'pitch' to the lower, or higher of the unison pitches per note.

It can give fine change to aspects of the way the whole piano tuning behaves and can improve the fidelity in crowded music (with many notes/chords, high velocities esp. bass harmonies etc.). Or it can just seem to make a piano sweeter or a little more sour, if that's the goal.

When straight up A/B testing, some chords might show little has changed - but playing something with lots of action (or listening in the MIDI player) can help - replay/listen/try new number.

I've noticed so far, the settings I mostly use for unison balance, are numbers between -0.2 and +0.4 - interested to know if others enjoy results from adjusting this control and other observations?

I'd be curious to know ...
<...>
Do other pianos--especially sampled ones--have a better level of tuning accuracy than PTQ?

Please note:  I am here only to report my real-life experience with poorly tuned sampled pianos.  This is not intended as standing on a soapbox to bash sampled piano libraries as a whole.

Long Answer:
One of the many reasons I permanently deleted all sampled high-end piano libraries (including but not limited to EWQL, Garritan, Synthogy, Dan Dean, Native Instruments, etc.) from my computer had to do with poor intonation in the commercially released versions.  Certain velocity ranges of individual notes -- and not always the same notes -- exhibited such poor mis-tuned trichords (clearly audible by ear) that I could not understand how a commercial product was allowed to be released to the public with these tuning flaws. 

In retrospect, when a software developer samples a piano in multiple velocity layers, it is inevitable that some combinations of notes and velocities will have gone out of tune, especially in slipping of individual trichords.

I distinctly recall speaking in person with the actual creator of a very-well recognized sample library at the 2009 Winter NAMM show, asking how such a product could be released with so many tuning defects in each brand of piano they had sampled.  His paraphrased reply went something like this:  "Well, we did re-tune the piano twice a day while sampling it."  My conjecture is that the sampling process was somehow automated, with up to eighteen velocity layers per note sampled.  During this time, it would be easy for at least some of the note-velocity combinations to be sampled without anyone actually listening to them.  Then upon beta testing, with so many notes and so many velocity layers, I can understand how some mis-tuned note-velocity combinations might slip through their quality control process, although there's still no excuse for these flaws to exist after extensive beta testing.

The way I used to work with sampled piano libraries would be to play a given piece, live, while recording a backup copy to midi, and then reviewing the midi file afterward in a DAW (usually Digital Performer Version 5 or Logic Pro Version 6 at the time).  Whenever I would find notes that sounded distinctly out-of-tune, I had to manually find the note in the DAW, and modify the note-on velocity to fire the next nearest same-note sample that was not still out of tune.

When I acquired Pianoteq version 3 in the first decade of the twenty first century, I felt terrifically relieved in NOT having to edit specific note-on velocities that happened to otherwise trip out-of-tune samples. 

Short answer:
Other pianos, including sampled pianos, do NOT have a better level of tuning accuracy than Pianoteq.

Cheers,

Joe

For informational purposes, at A440, a deviation of +4 cents results in a change of 1 hertz (i.e., 441Hz).  By this logic, at A440, a deviation of 1 cent results in the tuning being off by 1/4th Hertz (i.e., 440.25Hz), or one undulation of constructive/destructive interference for every four seconds the note is sounded.  When we deal with a theoretical accuracy of 0.1 cents, we're speaking of one-fortieth of a hertz @A440 (i.e., 440.025Hz).  Personally, I know of no human tuner who can ensure an accuracy of A440 +/- 1/40th Hertz by ear or with tuning hammer technique.

My two "cents" [pun intended].  This is a great discussion; please feel free to agree or disagree with my musings.

Cheers

Joe

I agree completely.  I doubt there could even be mechanized or laboratory conditions that would permit that across an entire instrument, let alone on an instrument that's going to be played.

I remember on my old upright, occasionally notes would slip out of tune with a week of tuning, just because the 1928 instrument's rusted strings and tired pinboard couldn't really take any slight adjustment--let alone a pitch raise.  No amount of passion for that piano kept it in tune.  There was one dichord--a D string, I think--that slipped significantly out of tune on a regular basis after maintenance (I think the string had snapped the year before and had been repaired).  The tech came back and tuned it a few days later, set the pin harder, and two days later it was out again.  He glued it, and two days later, the glue failed and it dropped tune, so finally he just found the strongest epoxy he could and glued it twice in two days, which happily worked: I don't know what outside of a rebuild would have been an option if it hadn't because you can't do that much to "reset" a pin without replacing all of them.  There are just factors in the real world, like that notorious string, that put practical boundaries on perfection.

Mathematical tuning just doesn't exist for pianos, or if it does, it's gone in seconds.

I'd be curious to know from the OP, if there is a particular need for having less than 1 cent accuracy for a project?  Is it because it will pair with synthesized or electronic instruments and the slightest beating would be a problem--where an e-piano or similar substitute might serve better?  Do other pianos--especially sampled ones--have a better level of tuning accuracy than PTQ?  Do the tuning app and microphone and other equipment used in this test have an guaranteed level of accuracy?

My two "cents" [pun intended].  This is a great discussion; please feel free to agree or disagree with my musings.

Cheers

Joe

Excellent post! I believe there is no such thing as a "perfectly in tune" piano. And even if there was, it would not necessarily be a good thing!

I first noticed this thread when it was posted on 12JAN2020, and am not sure the following question has been adequately stated:  "Is Pianoteq out of tune because it "sounds" out of tune or because it "measures" out of tune?"

I ask this because I feel that Pianoteq 6 PRO instrument presets sound completely in tune to my sense of hearing, right out of the box.  I am not terribly concerned about whether a given Pianoteq preset measures within a few cents of mathematical perfection for equal temperament.  Why not?  That's because when I tune any real acoustic piano to perfect equal temperament with a false sense of security by using an electronic tuning device, the piano is technically "in tune"; however, I feel that I can make the piano 'sing' more when I slightly deviate from theoretically/mathematically perfect tuning. 

One aspect of what I say has to do with the tuning of trichords (notes with three steel strings per key):   I strive to tune the central steel string to the best of my ability, then I generally tune the leftmost string with respect to the center string to emphasize even harmonics (octave harmonics), and I generally tune the rightmost string to the center string such to emphasize the odd harmonics (fifths, major thirds and major sixths, etc.).  Then I usually go back and make sure that the center string remains in as perfect tuning as I am capable of achieving. 

To repeat, each acoustic piano I tune professionally has certain deviations from mathematical perfection of equal temperament (meaning each semitone's fundamental frequency varies by the twelfth root of 2, or 1.05946309 times the reference tone's fundamental frequency).  I am wondering if, by noticing these slight mathematically theoretical variations in tuning, the Pianoteq presets "measure" out of tune.

For informational purposes, at A440, a deviation of +4 cents results in a change of 1 hertz (i.e., 441Hz).  By this logic, at A440, a deviation of 1 cent results in the tuning being off by 1/4th Hertz (i.e., 440.25Hz), or one undulation of constructive/destructive interference for every four seconds the note is sounded.  When we deal with a theoretical accuracy of 0.1 cents, we're speaking of one-fortieth of a hertz @A440 (i.e., 440.025Hz).  Personally, I know of no human tuner who can ensure an accuracy of A440 +/- 1/40th Hertz by ear or with tuning hammer technique.

My two "cents" [pun intended].  This is a great discussion; please feel free to agree or disagree with my musings.

Cheers

Joe

May I make a suggestion. Try this. Pick the Default Blüthner BA Solo Recording Preset with no adjustments. Now adjust the Stereo Width slider to read 401. Now play a chromatic run and let me know if there is any improvement.

I've done the above (with velocity curve alterations for my Yamaha P515) and am very pleased with the result. The setting of Stereo Width 401 works well with the Default Bechstein Sweet Preset also. I haven't tried any others yet but intend to tomorrow.

Let me know what you find,

Warmest regards,

Chris

That's not to say that you shouldn't look into the various tuning possibilities suggested! I'm a fully qualified pianotuner myself so understand the enjoyment that can be had when making a piano sound just how you want it. But for now you may prefer a simpler solution if indeed you find it works for you.

I've done the above (with velocity curve alterations for my Yamaha P515) and am very pleased with the result. The setting of Stereo Width 401 works well with the Default Bechstein Sweet Preset also. I haven't tried any others yet but intend to tomorrow.

Let me know what you find,

Warmest regards,

Chris