What's new

Which temperament to use for a virtual piano to play with strings libraries ?

darkneo57

Member
Hi,

I am having problems with notes that are out of tune in CSS & CSSS. I'm not the only one to notice this.

However, I just saw that the temperament of my virtual piano is set to "stretched" by default, I think I should set the temperament of the virtual piano to "equal" to make it sound more accurate with the library strings.

Can someone confirm that except in solo piano, when you want to make a piano and strings or/and brass (orchestral) project, you have to set the temperament to equal?

What seems strange to me is that by default the virtual pianos UNA CORDA and NOIRE are set to "stretch".


Thank you very much for your help
 
They are set that way because that's the standard today for keyboard instrument in order to compensate for the inharmonicity in the piano tuning. Even in piano concertos the discrepancy between the piano stretched tuning allows it to pop out in more complex passages and stand as a soloist (that's even more true when it is used in the orchestral palette as a percussion instrument).
However it might be desirable to select Equal temperament if you don't want that effect and you're using it like some sort of pad layering its sound with something else. But even then it is a hard call because it sounds a little unnatural.
 
Piano tuner here.

"Stretched" isn't a temperament; it's a necessary condition for a piano to sound in tune, because inharmonicity makes the piano's partials further apart than whole number ratios. Every (tuned) piano you have ever heard in your life is stretched. The question is by how much. There, the physics of the specific piano model plus to a smaller degree, the taste of the piano tuner (mostly in the bass and high treble) come into play.

The octave between the A below middle C and the A above middle C (the traditional starting octave for piano tuning) is stretched so that the 4th partial of the lower A lines up pretty well with the 2nd partial of the higher A (well, actually, the 4th partial is usually lowered to be just a smidgen wide of pure). That yields a good stretched octave for that piano (the octave will sound about as in tune as it can to a musician's ear). Then, in equal temperament, that octave is divided into 12 equal steps. That's what makes it equal temperament. (There are other schemes, such as a pure 3:1 P12th divided into 19 equal steps, but the first scenario is overwhelmingly the most common.)

Orchestras play with pianos with stretched tunings all the time. It works, because orchestral musicians in the treble area would rather be sharp than flat, so they tend to "juice" their octaves a bit, especially to match the piano. And in a piano concerto, if the piano treble stands out a bit, that's often a good thing. In the lower area, violists and cellists often tune their open strings to pure P5ths, (which are then wider than an equal tempered fifth), so that builds a bit of stretch into the orchestra (that idea doesn't hold for the basses, whose open strings 4ths in equal temperament are wider than pure P4ths, but I've never heard that cause a problem).

The only time I might adjust treble piano tuning for less treble stretch in an ensemble situation would be when the literature includes the piano playing in unison with tuned treble percussion that cannot adjust their tuning to match the piano (xylophone, etc). I rarely have to do that.

Not sure what VI you're referring to, but the terminology is suspect. If you try to tune a VI piano with 2:1 frequency ratio octaves, it will sound terribly out of tune. I suspect that the maker might mean "less stretched at the extremes" when they say "equal."

Pianoteq makes it easy to explore both stretch and temperaments (equal and tons of variants), if you want to experiment.
 
Last edited:
Joe explained this better than I ever could, but let me just add that if CSS is out of tune, the best solution probably isn’t to mess with the pianos stretched tuning but to use pitch correction on the notes in CSS that are out of tune. Or to just accept that the sampled strings probably aren’t worse than a real orchestra would be. At least it can’t be as bad as LASS? :)
 
thank you for your answers.

I'm sorry, but the information I found on the internet all says that the current pianos are tuned with the "equal" temperament. ( see wikipedia link and other )

https://en.wikipedia.org/wiki/Piano_tuning#:~:text=Pianos are usually tuned to,note A above middle C.

https://www.oudforguitarists.com/pianos-are-out-of-tune-equal-temperament-oud-music/

I forgot to mention that I'm a piano teacher, I knew that it was necessary to "cheat" for tuning, but I never went further than that. ( shame on me ). When I turn on my Roland RD 700 NX piano, by default the system is on "equal" 5 (see pictures attached)


roland.png

I'm not trying to contradict you, I'm just trying to understand, I found the same results by searching in French, but I find it really strange that by default the Vi piano I have are on "stretcheds".) ( see picture )

NOIRE.png

UNA CORDA.png

Could you please tell me why I can't find the same information as you please.

Thank you for your help
 
If you open up the manual or the Info in kontakt (F9) for all Galaxy pianos (including una corda and noire) under the stretched tuning section you will find:
Temperament Selector: Switches the tuning (or temperament) of UNA CORDA between Stretched and Equal. The default setting is Stretched, which is the way the piano was tuned for sampling. Stretched tuning accommodates the natural dissonances of metal strings, which stretch harmonics beyond the strings’ ideal frequencies. To solve this issue in order to make the instrument sound more harmonically pleasing (or “musical“), higher notes are stretched upwards. The amount of stretching required for a perfectly musical tuning depends on the length of the string, and thus the size of the piano.

But stretched isn't a temperament/tuning on itself as Joe_D has arleady pointed out. A stretched tuning is still equal temperament + you line up the upper partials of lower strings to the high strings. This second process makes the tuning go slightly sharper after midlle C and viceversa slightly flat towards the bass strings and makes the overall instrument sound better because of this relation established between the two extremes of the keyboard.
I think this box is actually mislabeled because in previous Galaxy Pianos (like Vintage D, the Vienna Grand, etc.) you could also choose Non equal temperaments like Werckmeister, Valotti, Silbermann... In most recent ones they have been removed and you are left with just stretched or not stretched equal temperament. So you're not changing the tuning by selecting one, just choosing between stretched (natural recordings) or not stretched.
 
darkneo57, the makers of the Galaxy piano have made a category error, and you are also making the same error.

By category error, I mean that it only makes sense to offer as options things that are in the same category. If someone asks you "is that thing in the corner black or is it a cat?", they are making a category error. It could be both black and a cat, or black and a dog, or gray and a cat, or black and a dog, etc. The question doesn't really make sense.

Stretch is one category. It refers to how far apart two notes that make up an octave are (or double octave, etc). An organ, for instance, generally would not have any stretch. Its intervals are harmonic, making tuning octaves and other big perfect intervals easy; they are simply tuned to whole number frequency ratios (octave A's are tuned at 55 Hz, 110 Hz, 220 Hz, 440 Hz, 880 Hz, etc).

As I said above, all piano tunings are stretched, meaning that the octaves are larger than whole number ratios. If the first note to be tuned is the A above middle C set to 440 Hz, The A below it will be lower than 220 Hz, and the A above it will be higher than 880 Hz. The only question is how much lower or higher, called the amount of stretch, and that is decided by the physics of the piano and to a lesser degree, the taste of the tuner.

Temperament is a different category. It refers to how the octave is divided into 12 steps. If all steps are equally distant from each other, we call that equal temperament. Playing a perfect fifths chromatically (C-G, C#-G#, D-A, etc), all fifths should sound almost exactly the same. Playing major thirds chromatically (C-E, Db-F, D-F#, etc), all of them should sound extremely uniform, though if you listen really carefully, the M3rds will have a subtle beating two octaves above the higher note (at the 5:4 partial coincidence) that will very slightly but uniformly get faster as you ascend chromatically. That's equal temperament.

Other temperaments divide the octave into unequal steps. An early temperament that is very unequal is just intonation. I would not recommend it for general use. In just intonation, the fifths in the primary triads (I, IV, V), are tuned pure (3:2 ratio, more pure than equal temperament), and the thirds in those same chords are tuned pure (which makes them much lower than equal temperament, sounding strange to our "modern" ears), and the remaining notes are tuned to compromises. If you play ascending chromatic P5ths, you will hear that some fifths sound great, others sound absolutely horrible! Likewise, if you play ascending chromatic M3rds, they will all sound very different from their neighbors -- some will be "sweet" while others will sound discordant. If you play a piece in the key in which the temperament was tuned, it will sound very consonant. If you transpose the piece to another key, it may sound very dissonant, or a mix of consonant and dissonant, depending on which other key you choose.

Just intonation is just one of hundreds of unequal temperaments. Most of the other temperaments are somewhere between equal temperament and just intonation, but they all have unequal interval sizes. They are named, often after the person who first wrote about them (Vallotti, Young, Werckmeister, etc). The "later" temperaments, such as those that J. S. Bach may have used, are closer to equal, so there is a different degree of consonance in various keys, but none of them sound horrible, and none of them sound as "sweet" as just intonation.

Your sources are largely correct in saying that in modern Western music, equal temperament is almost always used in tuning pianos. But I know an excellent tuner who tunes a very slightly unequal temperament, and his clients, including some recording studios, are happy with it. By coincidence, I will be tuning an unequal temperament tomorrow for a client who requested it. But it is rare, except in early music period practice performance, "new" music, and world music (where there are many tuning systems that don't really fit into Western temperament ideas).

String quartets and the best choirs (especially straight-tone Renaissance choirs) often adjust their intonation scheme as they perform, and may choose to play important chords closer to just intonation. They might play an E at one pitch in a C major chord, but play an E in an E dominant seventh chord at a slightly different pitch. Since they perform without frets or other devices which fix the pitch of their notes, they can adjust as they go, and it sometimes consumes a lot of their rehearsal time.

So, to sum up, in pianos, first an amount of stretch (for the octaves) is chosen, then the octave is divided into 12 smaller intervals called a temperament. If those 12 notes are spaced equally it's equal temperament. If they are spaced unequally, it is some other temperament, which probably has a name. And the makers of Galaxy pianos are causing confusion by making a category error.
 
So, to sum up, in pianos, first an amount of stretch (for the octaves) is chosen, then the octave is divided into 12 smaller intervals called a temperament. If those 12 notes are spaced equally it's equal temperament.
I've always wondered about that - so "equal" temperament means equal only within a single octave?

I would have thought there's some kind of algorithm used to distribute the variation across, say, groups of two or three octaves where they're windowed over each other with an octave overlap or something like that. It seems like treating each octave independently would sound more out of tune than some kind of multi-octave averaging. But I guess not!

Is there a standard for which octaves are used as the anchors? C to C? I guess you could use any octave pair and adjust the few stragglers at the top or bottom.

Interesting info - thanks!

rgames
 
I've always wondered about that - so "equal" temperament means equal only within a single octave?
Traditionally, an A to A (sometimes C to C) octave is set (stretched so that it sounds good). Then, that octave is divided into 12 equal steps. Then the notes within that “temperament octave” are used to expand the tuning up and down. For instance, using an A temperament, once the 12 notes are set, the tuning is expanded downward to add notes as follows: the G# in the temperament octave is used to tune the G# one octave below it, which is the note one half step below the lowest note in the temperament octave. The size (stretch) of this new G# octave is carefully controlled by using what are known as aural tests and checks, which are faster beating intervals like 6ths and 3rds. These tests and checks give the tuner confirmation that their octave is exactly the same size as the original A-A octave, and also that their “new” G#-A half step is exactly the same size as the adjacent A-A# half step. The process then continues for the G octave, and so on.

To your point about subdividing a larger span being potentially more accurate, there is a two octave temperament scheme that I like. But a skilled tuner will arrive at pretty close to identical results using whatever scheme they pick for a given tuning; aural tests and checks allow us to “prove” that our work is consistent.

The half steps should remain more or less the same size for the whole piano. Any section should have practically identically sized half steps within that section. There is more octave stretch in the high treble and low bass, so technically those half steps would be a tiny bit bigger. But, again, they should be the same size as their neighboring notes.
 
Traditionally, an A to A (sometimes C to C) octave is set (stretched so that it sounds good). Then, that octave is divided into 12 equal steps. Then the notes within that “temperament octave” are used to expand the tuning up and down.
Can you simply use a tuner and just go key by key? Do you have to do it by ear?
 
If you have professional-level piano tuning software, then yes, you can do it that way. Otherwise, no.

A guitar tuner type app won't work for tuning a piano, because it won't take into account the inharmonicity of a piano. It will instead usually use whole-number ratio tunings (220 Hz, 440 Hz, 880 Hz, etc). It would probably work for tuning an organ, though.

Every piano model's inharmonicity is different, and that inharmonicity isn't the same across any piano either. Inharmonicity gets higher as the treble ascends, usually has an upward spike in the low tenor plain steel strings, and often rises in the low bass. To tune a piano successfully, you (or the app) have to be able to hear or measure the inharmonicity across that specific piano, and adjust your octave width so that the octaves (and double octaves, triple, etc) will sound as good as they can. So, every model piano is tuned differently. The A above middle C is the typical starting point, it is usually set to 440 Hz, but after that, the other notes are put at a very slightly or wildly different frequencies on different pianos. The notes around the starting A will be tuned extremely close to the same way on different pianos, but the further you go from that starting area, the more differently the same note will be tuned on different pianos.

Professional piano tuning software measures and accounts for the inharmonicity of every specific piano, and allows the tuner to incorporate their taste in stretch as well. It is quite sophisticated. It's also very expensive, so it's not something folks typically buy just to play around with.

If you do tune using software, it is still best to use aural tests and checks as you go; the end goal is for it to sound tasteful and musically pleasing to the human ear, so using the ear to verify (and adjust if necessary) is still very important, IMO.
 
If you do tune using software, it is still best to use aural tests and checks as you go; the end goal is for it to sound tasteful and musically pleasing to the human ear, so using the ear to verify (and adjust if necessary) is still very important, IMO.

I once had a conversation about this with my tuner (who did not use software) and he said he always tunes in a way that is in tune and emotionally satisfying.

**Leigh
 
I am a professional
Piano tuner here.

"Stretched" isn't a temperament; it's a necessary condition for a piano to sound in tune, because inharmonicity makes the piano's partials further apart than whole number ratios. Every (tuned) piano you have ever heard in your life is stretched. The question is by how much. There, the physics of the specific piano model plus to a smaller degree, the taste of the piano tuner (mostly in the bass and high treble) come into play.

The octave between the A below middle C and the A above middle C (the traditional starting octave for piano tuning) is stretched so that the 4th partial of the lower A lines up pretty well with the 2nd partial of the higher A (well, actually, the 4th partial is usually lowered to be just a smidgen wide of pure). That yields a good stretched octave for that piano (the octave will sound about as in tune as it can to a musician's ear). Then, in equal temperament, that octave is divided into 12 equal steps. That's what makes it equal temperament. (There are other schemes, such as a pure 3:1 P12th divided into 19 equal steps, but the first scenario is overwhelmingly the most common.)

Orchestras play with pianos with stretched tunings all the time. It works, because orchestral musicians in the treble area would rather be sharp than flat, so they tend to "juice" their octaves a bit, especially to match the piano. And in a piano concerto, if the piano treble stands out a bit, that's often a good thing. In the lower area, violists and cellists often tune their open strings to pure P5ths, (which are then wider than an equal tempered fifth), so that builds a bit of stretch into the orchestra (that idea doesn't hold for the basses, whose open strings 4ths in equal temperament are wider than pure P4ths, but I've never heard that cause a problem).

The only time I might adjust treble piano tuning for less treble stretch in an ensemble situation would be when the literature includes the piano playing in unison with tuned treble percussion that cannot adjust their tuning to match the piano (xylophone, etc). I rarely have to do that.

Not sure what VI you're referring to, but the terminology is suspect. If you try to tune a VI piano with 2:1 frequency ratio octaves, it will sound terribly out of tune. I suspect that the maker might mean "less stretched at the extremes" when they say "equal."

Pianoteq makes it easy to explore both stretch and temperaments (equal and tons of variants), if you want to experiment.
Very good explanation. I am a professional tuner ( by ear) myself and could not have explained it better. Therefor a concert with two grands, besides of volume difference, needs two grands preferable with similar string lengths when two identical grands are not available. The streched octaves on a small upright are giving you a different result in the upper and lowest octaves compared to a concert grand. There is where a tuning device fails miserably ! Tuners who tune a small cabinet upright with extremely short bass strings with inaccurate fundamentals will always be tuned to ‘high’ when using a tuning device. Even when it’s impossible to tune a non beating octave with such a short scaled bass your ear gives you the ‘only way out’ to get the ‘best‘ results, whereas a tuning devise gives you an ‘all over the place’ result and way to ‘high’ projected bass register for any desirable outcome.
 
Can someone confirm that except in solo piano, when you want to make a piano and strings or/and brass (orchestral) project, you have to set the temperament to equal?
Regardless of the temperament you use and regardless of a stretched tuning, a piano will never really blend with strings.
A piano inside of strings always has its distinct sound that sets it apart, and this special sound can be amazing or not so amazing.

A piano playing arpeggiated chords that are held by the string section sound beautiful if it happens in the context of a piano concerto (or a solistic part) where you still want the piano being an individual voice.

However, a piano playing arpeggiated chords in other contexts is often just bad orchestration (and a mistake permanently found). Often, a harp is a much better choice. Or arpeggios played by the (bass-)clarinet. Etc.

A piano can be a good addition to string lines if it's about creating a percussive impact (where the metallic overtones can stand out without problems). With soft lines, the combination often sounds awkward.

However, the tuning of the piano doesn't change any of these aspects.
 
Top Bottom