Inharmonicity of sample piano sounds
Steve Freides
real, acoustic piano sounds? I just assumed that they would, and that
therefore the octaves would be somewhat stretched, again to mimic an
actual piano.
-S-
Elrond Hubbard
@mid.individual.net:
If they're sampled from a real piano, then, well, wouldn't they?
EH
Steve Freides
I agree, but someone on another newgroup suggested that, because the
sounds are electronic in nature, they wouldn't.
I asked here because, although we are all well able to speculate, I was
hoping someone might actually have a fact-based answer, perhaps someone
who was involved with the design and manufacture of these things or
otherwise had some specific knowledge to share. I know what sampling is
in principle, but I'm also well aware that, once sampled, sounds can be
manipulated in the digital domain before being "put" into the
instrument. Perhaps they take the inharmonicity out in part or in
whole? I have no idea.
Nice forum handle, BTW.
I. P. Daily
Fred Williams
Now I always thought that the sampling refer to the waveform that gives
each note it's tone, but the pitch was achieved by mathematical means
afterwards, which means they can set whatever temperament they like.
--
Regards,
Fred
(remove FFFf from my email address to reply by email)
Steve Freides
Right, but now we're back to talking about what inharmonicity is and how
it matters to tuning. In a nutshell, if the second partial of a
string/key that's tuned to 100 Hz is not 200 but 202, then you would
tune the higher octave to 202 and not 200. It's not an issue of
temperament but of tuning to match the slightly out of tune harmonics of
the strings you have. My example is actually backwards of the way it
usually works, e.g., you tune the middle of the piano, and you have a
string at 200, and then when you find the second partial of your next
octave down is 202 instead of 200, you stretch the octave, tuning the
lower string to !99 to that its second partial is in tune to the
fundamental of the note an octave above.
-S-
Elrond Hubbard
>>> I asked here because, although we are all well able to speculate, I
>>> was hoping someone might actually have a fact-based answer, perhaps
>>> someone who was involved with the design and manufacture of these
>>> things or otherwise had some specific knowledge to share. I know
>>> what sampling is in principle, but I'm also well aware that, once
>>> sampled, sounds can be manipulated in the digital domain before
>>> being "put" into the instrument. Perhaps they take the
>>> inharmonicity out in part or in whole? I have no idea.
>>>
>>> Nice forum handle, BTW.
Thanks.
>> Now I always thought that the sampling refer to the waveform that
>> gives each note it's tone, but the pitch was achieved by mathematical
>> means afterwards, which means they can set whatever temperament they
>> like.
My understanding of sampling is that you are making a faithful recording
of the complex waveforms that make up the sound you want to reproduce.
So,
in the case of the piano, the sample would reflect any inharmonicity in
the piano being sampled.
>
> Right, but now we're back to talking about what inharmonicity is and
> how it matters to tuning. In a nutshell, if the second partial of a
> string/key that's tuned to 100 Hz is not 200 but 202, then you would
> tune the higher octave to 202 and not 200. It's not an issue of
> temperament but of tuning to match the slightly out of tune harmonics
> of the strings you have. My example is actually backwards of the way
> it usually works, e.g., you tune the middle of the piano, and you have
> a string at 200, and then when you find the second partial of your
> next octave down is 202 instead of 200, you stretch the octave, tuning
> the lower string to !99 to that its second partial is in tune to the
> fundamental of the note an octave above.
Right. Which is why manufacturers sample concert grands for their
electronic keyboards - longer strings having less inharmonicity. I
suppose there's a place for a sampled Baldwin Acrosonic, but it it isn't
a place I want to visit.
EH
Steve Freides
Which brings us back to my original question - is the inharmonicity of a
concert grand preserved in, e.g., the electronic keyboard I happen to
own, a Yamaha P85, or other similar keyboard? I agree that it sounds
like it should be preserved, I'd just like to get a fact, not an
opinion, if possible.
FWIW, my "other" keyboard is a Del Fandrich rebuilt 1913 Knabe 5' 2"
grand, a delightful instrument in every way.
-S-
Laurence Payne
wrote:
>>> Now I always thought that the sampling refer to the waveform that
>>> gives each note it's tone, but the pitch was achieved by mathematical
>>> means afterwards, which means they can set whatever temperament they
>>> like.
>
>My understanding of sampling is that you are making a faithful recording
>of the complex waveforms that make up the sound you want to reproduce.
>So,
>in the case of the piano, the sample would reflect any inharmonicity in
>the piano being sampled.
There's two aspects to the sampling, pitch and volume/timbre/attack.
Maybe every note, all 88, are recorded and played back at their
original pitch. Maybe (say) every third note is recorded and the gaps
are filled by adjusting playback rate. Maybe the playback map can also
be micro-adjusted to allow different tuning schemes.
On all but the cheapest instruments, each sample will also be taken at
varying volume levels, from a softly struck key to a fortissimo attack
(which differ in may ways other than simply volume). As key velocity
increases, the sampler cross-fades between this choice of samples.
The sound designer also has to decide what to do when the damper pedal
is used - whether to settle for simple sustain or to attempt to
emulate the resonance of unplayed strings.
Just like the design and construction of a real piano, a lot of this
is science, some is art. Go in with measuring instruments and an
analytical ear - it's easy to find faults (but it is on a Steinway as
well :-)
polymod
"Steve Freides" <st...@kbnj.com> wrote in message
news:7n8ubiF...@mid.individual.net...
This piano tuner got a good laugh out of that last line!!
Poly
Steve Freides
That all makes sense to me. I notice that some Yamaha electronic
keyboards now support things like the halfway position of the una corda
pedal.
FWIW, I emailed Yamaha, via their web site, and asked specifically about
the preservation or lack thereof of the inharmonicity present in the
samples in the "finished" electronic keyboard. I will post their reply,
assuming it's of any substance, when (and if) I receive it.
-S-
Elrond Hubbard
@mid.individual.net:
>>
>> Just like the design and construction of a real piano, a lot of this
>> is science, some is art. Go in with measuring instruments and an
>> analytical ear - it's easy to find faults (but it is on a Steinway as
>> well :-)
>
> That all makes sense to me. I notice that some Yamaha electronic
> keyboards now support things like the halfway position of the una corda
> pedal.
>
> FWIW, I emailed Yamaha, via their web site, and asked specifically about
> the preservation or lack thereof of the inharmonicity present in the
> samples in the "finished" electronic keyboard. I will post their reply,
> assuming it's of any substance, when (and if) I receive it.
I've often wondered if the technology exists to make a "perfect" piano
electronically - one where all of the partials of any note are perfect
mulitiples of their fundamental (in other words, no inharmonicity); where
each ascending partial's contribution to a note is in an evenly decreasing
proportion; and where playing a note excites partials in undamped strings
in proprtion to the volume of the note.
I don't know doodly about what is involved in modeling these things. But
I'm curious what it would sound like. There would be no need to stretch
the tuning, and every note would be at its theoretical value.
Any thoughts?
Jean-David Beyer
>
> I've often wondered if the technology exists to make a "perfect" piano
> electronically - one where all of the partials of any note are perfect
> mulitiples of their fundamental (in other words, no inharmonicity); where
> each ascending partial's contribution to a note is in an evenly decreasing
> proportion; and where playing a note excites partials in undamped strings
> in proprtion to the volume of the note.
I imagine I would not like it. If things like that were even desirable
they could probably be done now, but they would sound "wrong." Consider
bells. There, the upper partials are pretty much non harmonic. Some
churches have electronic bells that are not sampled, but have tuned
resonators (like the "comb" of a music box whose upper partials are more
harmonically related than real bells, and they sound terrible.
In a similar vein, one could make a perfectly tempered electronic piano,
where the computer part would recognize the key you were playing at any
instant and adjust the note's pitch accordingly. Keeping C# and Db
different in pitch, for example. Which might be all very well if one is
playing in only one key at a time. I do not know what it would do for
Charles Ives or Conlon Nancarrow or even Henry Cowell.
>
> I don't know doodly about what is involved in modeling these things. But
> I'm curious what it would sound like. There would be no need to stretch
> the tuning, and every note would be at its theoretical value.
>
> Any thoughts?
--
.~. Jean-David Beyer Registered Linux User 85642.
/V\ PGP-Key: 9A2FC99A Registered Machine 241939.
/( )\ Shrewsbury, New Jersey http://counter.li.org
^^-^^ 15:45:01 up 20 days, 22:19, 3 users, load average: 4.37, 4.32, 4.31
Laurence Payne
wrote:
>I've often wondered if the technology exists to make a "perfect" piano
>electronically - one where all of the partials of any note are perfect
>mulitiples of their fundamental (in other words, no inharmonicity); where
>each ascending partial's contribution to a note is in an evenly decreasing
>proportion; and where playing a note excites partials in undamped strings
>in proprtion to the volume of the note.
>
>I don't know doodly about what is involved in modeling these things. But
>I'm curious what it would sound like. There would be no need to stretch
>the tuning, and every note would be at its theoretical value.
There are other reasons to stretch-tune. Also remember that, on a
piano tuned in equal temperament, only octaves align perfectly with
the theoretical harmonic series.
No real instrument is perfectly in tune. But the ear accepts a
slightly-out-of-tune octave as concordant, even though the frequency
ratio is far more irrational than a perfectly-in-tune major 7th. Funny
thing, psycho-acoustics :-)
Steve Freides
You're mixing different things here. It is certainly possible to remove
inharmonicity electronically, and my guess is that none of us would mind
the sound of that at all. Just my untested opinion here, but I don't
think an overtone of 202 on top of a fundamental of 100 sounds any
better or worse than a perfectly in tune 200 does. What we like is to
perceive that the instrument is in tune with itself, especially as far
as octaves, fourths and fifth go.
But as to your next suggestion, having the partials gradually lessen in
some linear way - well, _what_ linear way? There are plenty of
instruments where the way the partials lessen is part of their
characteristic sound, and following a linear model would make them sound
worse, or at least less like the instrument they were supposed to be.
Attacks are also critical to the sound of an instrument - listen to a
recording of a piano played backwards as but one example.
By now, someone has undoubtedly tried all of what you're suggesting.
-S-
Steve Freides
and therefore a lot is done with a few, simple samples, e.g., no
sampling at different velocities. But they didn't address the issue of
inharmonicity, so I asked them again. Their full reply is below.
-S-
***************
The P85 specifically will not feature as much harmonic content as some
of our higher end models. The reason is that the P85 does not offer
individual sampling for each key or multi level sampling for velocity
triggering. Accurately capturing and reproducing harmonic content
requires much more memory within the instrument and many more levels of
velocity triggering. The P85 does it's best to represent the sampled
acoustic piano used as it's base for the price point it is offered at.
Many of our mid and upper end models will feature full stereo dynamic
sampling, four to five layer velocity triggering, full string resonance
sampling and stereo sustain sampling. The core sample of the P85 is of
the CFIIIs conservatory grand but many of the nuances that make the
acoustic piano an acoustic piano have been filtered and simplified
within it's digital version.
*****************
Steve Freides
*****************
That's why the P85 uses multiple samples through out the octave range.
It isn't a single sample stretched across multiple octaves.
*****************
So I think it's fair to say that inharmonicity is preserved in sampled
electronic keyboards.
-S-
RVG
The only software I know that preserves the enharmonics is Pianoteq.
Works as a Midi VST instrument in any sequencer/arranger/etc.
--
Jazz up your life!
Jazzez-vous la vie!
http://rvgmusic.bandcamp.com/
http://rvgjazznstuff.jamendo.net/
"La premi�re arme de la R�sistance c'est l'information." Lucie Aubrac
andrew s
Does anybody have a really good piano soundfont or SFZ sample?
"Steve Freides" <st...@kbnj.com> wrote in message
news:7nko64F...@mid.individual.net...
Laurence Payne
wrote:
>Thank you for all of this information.
>Does anybody have a really good piano soundfont or SFZ sample?
I doubt it. We've got used to very sophisticated multi-layer piano
samples in today's keyboards, even quite inexpensive ones. Soundfonts
aren't really up to the job.