Topic: stretch tuning tech details?

I have some technical questions about stretch tuning.

1. What does the parameter represent?  You give possible values of 1.00, 1.10, 1.15, and 1.50 as well as allowing custom values, but you don't seem to say what these values mean.  Is this parameter related to the inharmonicity constant B?

2. Do you use different values of B in different ranges, depending on the nature of the strings?

3. What happens when an alternative tuning is used?  For example, how would the well-temperament preset interact with a natural stretch tuning setting?  Do you simply apply a standard Railsback curve to all notes, independently of the temperament?

Late addition:
4. Why is it that the harpsichord presets include stretching?  My understanding is that the inharmonicity of harpsichords is typically negligible.

Last edited by Unhandyandy (15-10-2018 21:33)

Re: stretch tuning tech details?

Nothing?  Don't any Pianoteq engineers read the forum?

Re: stretch tuning tech details?

Hi Unhandyandy,

I hate to see something missed - it happens - if I had more time I might have had a reply earlier.

If nobody has seen this thread, it may be very likely that what you ask is a fairly rare pupper.

Forum users can guess only how exactly Pianoteq handles this.

I'd suggest a question to support, since it's likely to be directed to the person/team who will know intimately what you'd like to learn.


My guesses and experiences:

1.
Fixed numbers with internal fixed settings - feel free to choose any numbers (up to 3 iirc) - they may not (or may) refer to something IRL - but listen, if you like, use. Support may give more info so you can get more technically refined with it if that's your goal. Can't say anything re inharm const B.

2.
The piano model resolution is per string and interactions across the deck etc. and such detail in different rages is catered for regarding other aspects - although to what detail stretch tuning affects an exact piano with exact stretch params? I find it marvelous to choose between "String Tension" and "Full Rebuild" - taking into account how your tuning would sound if your current piano is stretch tuned (various inharmonicity effects - more with less tension - less with more tension), or a all-new 'version' of your piano is 'built' from scratch with the lengths for strings (to keep current timbres - per note and across the deck) without that extra strain on the frame etc. Tuning can sound better one way or the other. Not exactly what you asked, but provides a probable assumption that yes, many small things taken into account.

3.
Given the above, I'd be confident that whatever it does, it would be 'quite like' what might occur in reality if you took a piano and did what you describe - technically would require support to give detail. I find it perfectly wonderful with tunings + a little stretch. In the end, like a real piano and a skilled tuner (with producer and label all breathing down everybody's necks) it would come down to "Does this work? Is this my signature sound? Is it enough like that famous piano/recording/artist?". In the atomic world, nothing is perfect.. there's some next generation level of understanding this going on in the Pianoteq engine - which can be confusing because, on the one hand, you have set numbers for much of the control system, and yet, it's about how it all feeds through a 'real' piano + velocities + reverbs and so on, to create or recreate something 'of reality' in the virtual sense. It's not just samples and samples of harmonics being sprayed across the sound stage.

4.
Since tuning is "in" the engine, why not allow other instruments to have use of it, even if not normal (or realistically something you'd do IRL). You do not have to use it but it's interesting to try out "what if" scenarios certainly. Plus, many people use Pianoteq for more than realistic instruments, in fact in terms of modern music/musicians, it is remarkable for 're-synthesising' reality in these ways. Instead of traditional synthesis, it's a whole universe of 'reality based' elements which can (all?) be pushed beyond reality if that is to be our wont. Results can be astoundingly subtle (realism in mind) or entirely maniacal by virtue of the generous parameters.

Hope some of this helps.

Pianoteq Studio Bundle (Pro plus all instruments)  - Kawai MP11 digital piano - Yamaha HS8 monitors

Re: stretch tuning tech details?

As a piano tuner, I apply stretched tuning whenever I tune by ear the fundamental frequencies of the higher strings to match the overtone series of the lower-to-midrange strings. I do this on-the-fly without consulting a manual as to how much to raise the top octaves and lower the bottom octaves -- it is all done by ear.

As such, I tune in order to make the piano sound "natural" to my sense of hearing.  Restated, if one hears the higher notes to be out-of-tune with the lower notes' overtones ... by virtue of using an electronic tuner to dial in the so-called correct pitch of any given high note -- then by definition that high note will sound "flat" in pitch even though it is tuned independently of the other notes.

I apply only enough pitch stretching such that the constructive/destructive interference in two simultaneously sounded double- to triple octave notes has been eliminated.  Smaller pianos, including grands and uprights need more correction than do longer-lengthed concert grands.

Here is another way I would put this:  I can always tell when an amateur tuner (or relatively inexperienced professional tuner) has tuned all of the notes of a piano using only an electronic tuner set to equal temperament without listening to the job he did on the piano! EDIT:  In my opinion, if a piano tuner insists on using an electronic tuning device, then he should only use it to "lay the bearings" in the middle two octaves -- and then tune the rest by ear in octaves to the middle notes. End Edit Surely, such a piano tuner may protest by saying, "well, my tuning job matches what the electronic tuner described as being perfect".

Aside: Some tuners feel they need to apply an excess of stretched tuning because they believe it is "the thing to do".  This results in pianos whose highest octaves are over-tuned a quarter-tone to a semitone too high!  (Don't laugh; this occurs wayyyy too often in my experience!)  I brought this point up to a so-called "professional tuner" who retorted, "So what?  Nobody plays those high notes anyway!!"  I just walked away without arguing with that old fool.

* * * * * * *

Now, how does this apply to Pianoteq?  Well, since most of Modartt's pianos are concert grands with the exceptions of the Steinway B, U4 upright, and some of the smaller historic pianos, I believe the stretch tuning slider is more of a "cure without a disease", a nicety!  Why? try playing any two of the same notes approximately 3 octaves apart and listen for constructive/destructive interference in the overtones.  If you don't hear anything out of the ordinary (which I DON'T generally hear), then no stretching adjustments are required!

* * * * * *

Regarding harpsichord strings and the need (anti-need) for stretched tuning:  The strings of a harpsichord are extremely thin in comparison with corresponding notes of a piano.  Therefore, the end conditions of inharmonicity (the resultant boost of higher frequencies of the various overtone series) usually does not play a part in the sound of the instrument.  EDIT #2:  In addition, harpsichords do not boast 7-1/2 octave ranges of modern pianos; therefore, their limited number of octaves also preclude what little inharmonicity effect there is in their strings. End Edit #2 If you were to adjust the Pianoteq slider -- to the point that stretched tuning becomes noticeable to your ears when playing the same note 2 to 4 octaves apart -- then you have gone too far.  So, simply do not apply that amount of stretch to the tuning!

My two cents EDIT #3 (cents in terms of money rather than cents as a fraction of a semitone). End Edit #3

Joe

Last edited by jcfelice88keys (20-10-2018 04:41)

Re: stretch tuning tech details?

Unhandyandy wrote:

I have some technical questions about stretch tuning.

1. What does the parameter represent?  You give possible values of 1.00, 1.10, 1.15, and 1.50 as well as allowing custom values, but you don't seem to say what these values mean.  Is this parameter related to the inharmonicity constant B?

2. Do you use different values of B in different ranges, depending on the nature of the strings?

3. What happens when an alternative tuning is used?  For example, how would the well-temperament preset interact with a natural stretch tuning setting?  Do you simply apply a standard Railsback curve to all notes, independently of the temperament?

Late addition:
4. Why is it that the harpsichord presets include stretching?  My understanding is that the inharmonicity of harpsichords is typically negligible.


Ihnarmonicity is the property of strings that causes harmonics to differ from integer multiples of the fundamental.

1. Stretch tuning is what you use to counteract the effects of inharmonicity, and is the gradual sharpening (or flattening) the further you get from the keyboard centre. There is also a psychoacoustic phenomenon that changes pitch perception depending on the octave.

2. In a real piano yes, as there are particular issues caused by the thickness of bass strings, for example.

3. Stretch tuning is still used. This is something that tuners have been doing for a long time before they knew they were actually doing it.

4. The harpsichord also has some inharmonicity. Stretch tuning is used if it makes the instrument sound better. It's an aesthetic choice.

I was writing at the same time as the previous poster.

Last edited by johnstaf (17-10-2018 02:28)

Re: stretch tuning tech details?

Thanks for the replies.  The info from real piano tuners is especially valuable.

It's worth mentioning that the human ear may prefer intervals that are slightly wider than pure, even apart from any considerations of inharmonicity.

But I still want to know some of the mathematical details of how these concepts are implemented in Pianoteq.  I'll try emailing support.

Re: stretch tuning tech details?

Unhandyandy wrote:

It's worth mentioning that the human ear may prefer intervals that are slightly wider than pure, even apart from any considerations of inharmonicity.

I believe so. It's not just a case of lining up harmonics.

Re: stretch tuning tech details?

Hi Guys,

This maybe a little tangendental (is that even a word), so apologies in advance.

However, I have long been trying to get an even chromatic run that pleases my (piano-tuner) ears by first of all doing my best to make sure my keyboard's keybed is as level as I can get it, and then experimenting with the miriad of parameters available to me in the Pro version of Pianoteq. Tonight I began looking at octave stretching and am now very pleased with the results so far. I have Bluthner Recording BA selected as a preset, all note-by_note pianoteq default variations set to level, and an octave stretching value of 1.53 which I find gives me a very playable chromatic scale with minimum "standouts". I'm using a host (Cantabile 3 free version) which does not allow me to record but I will see if I can sort that so I can post a demo here. Meanwhile, I'd be grateful to anyone who may wish to try 1.53 octave stretching (440 equal temp.) To let me know what they (honestly) think.

Warmest Regards,

Chris

Last edited by sigasa (21-10-2018 22:09)

Re: stretch tuning tech details?

HI Sigasa,

If I understand this stuff correctly, the stretching factor should have no bearing on the evenness of a chromatic run.  That evenness should be determined solely by the temperament, which I guess you'd want to be ET.

Stretching should only affect the quality of intervals: octaves and to a lesser extend P5s and M3s.

Qualification:  Thinking about this more, I was wrong to say the evenness depends only on the temperament: corrections for inharmonicity could  affect evenness as well, if the inharmonicity varies over the keyboard (as it usually does).  But in that case stretching would only preserve evenness if it were set to 0.

On the other hand stretching could affect the quality of the semitones rather than their evenness.

Last edited by Unhandyandy (22-10-2018 20:40)

Re: stretch tuning tech details?

Hi sigasa,

sigasa wrote:

I'd be grateful to anyone who may wish to try 1.53 octave stretching (440 equal temp.) To let me know what they (honestly) think.

The honest simple answer is whatever you like is perfect! For public maybe a challenge.

I like it personally and find it very pleasing with a small fix.

On the plus side of stretching equal temperament to 1.53 is a surprising silken spread across harmonics. My ears confirm it's an interesting number! (the circular diagram in Pianteq's advanced tuning window is an interesting data point to observe). That amount of stretching comes at a small cost (to my ears) of a slight flatness - E inre C - esp. in 7th patterns which lose some brevity. It's subtle, took me a little while to adjust and discover what I like and don't like about it - like moving between historic tunings to feel out the subtleties.

In Pianoteq's advanced tuning window - I changed the E3 (cents listed on right side - click/edit) to 401, 402 or even 403 (borderline too much maybe) gladly it seems to very plainly correct whilst keeping the overall effect.

Other suggestion might be, to dial back some way from 1.53 & in advanced tuning window, select the stretch tab and select "Harmonic stretching". I find that adding harmonic stretching can reinforce some organic charm without higher stretch numbers, therefore not as much apparent effect on tuning per se. Also changing to "full rebuild" rather than "string tension" - may be worth trying (such an excellent extra, to hear these differences).

Also in that tab I really like "Transfer the stretching into Detune note-edit" - and viewing the curve - being able to tweak just one particular E for example to satisfy a piece's exact requirement quite super.

My preference for stretching equal temperament is around 1.08 as starting point (1.30s seem to be my instinctual cut-off point). I do begin to wonder how many others begin to hear the tuning as strange in some way - and always end up asking myself what you asked above - but hope my experiments help you in your observations - it was an enjoyable experiment, thank you sigasa.

There is pure mathematical data, but there's varied human experience subjected to this - it's a topic with no end of interesting ideas from historic tuning through to microtuning.

Pianoteq Studio Bundle (Pro plus all instruments)  - Kawai MP11 digital piano - Yamaha HS8 monitors

Re: stretch tuning tech details?

Thank you.

More experimenting. I ticked the harmonics- stretching option in micro tuning and am experimenting with lid positions. 0.76 is nice I think. I've set unison width and balance to level (1.00).

Warmest Regards,

Chris

Last edited by sigasa (23-10-2018 20:48)

Re: stretch tuning tech details?

Tonights offering;

Bluthner Concert Recording AB preset

Unison Width and Unison Balance levelled

Lid Position 0.71

Octave Stretching 0.99

Harmonic Stretching enabled

Opinions welcome,

Warmest Regards,

Chris

Re: stretch tuning tech details?

jcfelice88keys wrote:

As a piano tuner, I apply stretched tuning whenever I tune by ear the fundamental frequencies of the higher strings to match the overtone series of the lower-to-midrange strings. I do this on-the-fly without consulting a manual as to how much to raise the top octaves and lower the bottom octaves -- it is all done by ear.
Joe

Great explanation,  In regards to Pianoteq, I think a lot may be related to the particular speakers you are using and probably the sound card as well.  I know on the Model B I had to stretch the bass significantly to get it to sound "sweet" on my particular set up.  Prior to that, I hated the sound of that particular Pianoteq voice.  I have the pro version software, and did it note by note, not with a slider.

Pianoteq Pro 7.x - Kubuntu Linux 19.10 - Plasma Desktop - Hamburg Steinway

Re: stretch tuning tech details?

Unhandyandy wrote:

HI Sigasa,
Stretching should only affect the quality of intervals: octaves and to a lesser extend P5s and M3s.

On an acoustic piano the relationship between the P5s and M3s is totally critical.  The 5s must be shrunk for the M3s to sound sweet.  From my experience this is the primary battle in tuning a piano.  One thing I love about Pianoteq is the generally very beautiful tuning which is hard to achieve on an acoustic.

Pianoteq Pro 7.x - Kubuntu Linux 19.10 - Plasma Desktop - Hamburg Steinway