back to list

Tuning of samples re: jsz post

🔗Charles Lucy <makemicro@...>

1/24/2006 3:38:27 PM

My tuppence worth;

It seems to me that tuning your audio samples to 12edo referenced from 440Hz, would be the most convenient tuning for importation into other systems;
e.g. EXS24; Halion; Akai etc.

The system used for playback would then microtune then, and match other samples from other sample lbraries, as chosen from within the player software.
If you tune them to any other tuning, it will be necessary for the person importing them to tweak each sample mapping file for each sample to make the
sounded reference pitch be consistent with all other samples.

From my experience samples usually use between 12 and 3 samples per octave, dependent upon the audible changes in the timbres from note to note.

Charles Lucy - lucy@...
------------ Promoting global harmony through LucyTuning -------
for information on LucyTuning go to: http://www.lucytune.com
for LucyTuned Lullabies go to http://www.lullabies.co.uk
Buy/download/CD from: http://www.cdbaby.com/cd/lucytuned2

🔗Jon Szanto <jszanto@...>

1/24/2006 4:44:23 PM

Hi Charles,

{you wrote...}
>My tuppence worth;
>
>It seems to me that tuning your audio samples to 12edo referenced from 440Hz [snip] ... The system used for playback would then microtune then, and match
>other samples from other sample lbraries, as chosen from within the
>player software. If you tune them to any other tuning, it will be necessary for the person importing them ...

I suppose I might have prefaced my comments with my intended use, but this would be for personal use. I don't foresee anyone else using these samples. One of the reasons I was questioning Prent about this was how much re-sampling, if at all, he has done. There are some samples that are perfectly stretchable or compressable to just work with someone else's 12tet set. But certain timbres seem to really show a different charcter before they've gone 100 cents up or down, so I'm not necessarily ready to just bend things away from 12tet and be satisfied.

Anyway, this would all be for a purpose-driven project/composition, and I'd probably settle on a couple of tunings early on to design the sounds/instruments.

Cheers,
Jon

P.S. Thanks for the cc on the note!

🔗Rick McGowan <rick@...>

1/24/2006 5:40:07 PM

If we're talking blue-sky here about how samples *should* be stored and
tuned by samplers... Sample formats should not rely on the samples being
actually in-tune, even to 12-tet. The sample *format* should store the
actual nominal frequency of each sample, and the software that uses it
should properly tune it as needed, with reference to the specified
frequency. I think this is a big source of errors in existing sample sets,
as Jon can probably attest. If you start with a slightly out-of-tune
instrument, even by a small amount that's not really noticeable if used
"straight" in 12-tet it may sound *very* out of tune when you start
modifying its frequency (resampling) for use with other scales.

(Aside: Rhino apparently does the right thing. Though it's not a sampler,
it can import samples. Rhino also supports microtuning. In the Rhino input
format for samples, you are required to specify either the note ("c3") or
the actual frequency. I always specify the measured frequency, and as long
as it's correct, the tuning seems correct.)

Rick

Charls Lucy wrote,

> The system used for playback would then microtune then, and match
> other samples from other sample lbraries, as chosen from within the
> player software.

🔗Graham Breed <gbreed@...>

1/25/2006 3:23:43 AM

Rick McGowan wrote:
> If we're talking blue-sky here about how samples *should* be stored and > tuned by samplers... Sample formats should not rely on the samples being > actually in-tune, even to 12-tet. The sample *format* should store the > actual nominal frequency of each sample, and the software that uses it > should properly tune it as needed, with reference to the specified > frequency. I think this is a big source of errors in existing sample sets, > as Jon can probably attest. If you start with a slightly out-of-tune > instrument, even by a small amount that's not really noticeable if used > "straight" in 12-tet it may sound *very* out of tune when you start > modifying its frequency (resampling) for use with other scales.

Sound Fonts have a field for the actual frequency.

Graham

🔗Jon Szanto <jszanto@...>

1/25/2006 7:46:00 AM

Graham,

{you wrote...}
>Sound Fonts have a field for the actual frequency.

Yes, but (naturally) there aren't currently any supported, useable soundfont editors. The PD one's I've tried are terrible or crash, and original ones like Vienna require a SoundBlaster card. I'm waiting for a new version of Extreme Sample Convertor, due in about a month, that may allow for sample conversion and editing with this in mind.

Still haven't heard from Prent...

Cheers,
Jon

🔗Rick McGowan <rick@...>

1/25/2006 8:06:14 AM

Graham wrote...

> Sound Fonts have a field for the actual frequency.

Good news... do you know if people actually use it? ;-)
I haven't seen documentation on the soundfont format. Does anyone happen
to have a pointer to on-line documentation of it?

Thanks,
Rick

🔗Graham Breed <gbreed@...>

1/25/2006 1:15:46 PM

Rick McGowan wrote:
> Graham wrote...
> > >>Sound Fonts have a field for the actual frequency.
> > > Good news... do you know if people actually use it? ;-)
> I haven't seen documentation on the soundfont format. Does anyone happen > to have a pointer to on-line documentation of it?

I found a link from here:

http://soundblaster.com/soundfont/faqs/

You'll have to check the details. From what I remember, if you're using tuned samples (i.e. not drums) there's no other way to know what note the sample's for, and so it wouldn't make sense to ignore this field. The format itself is fairly widely used, the nearest to sample lingua-franca, but not easy to edit as Jon pointed out.

Graham

🔗Carl Lumma <ekin@...>

1/25/2006 5:46:31 PM

What about Reason, Halion, GigaStudio, Kontakt?

-Carl

At 07:46 AM 1/25/2006, you wrote:
>Graham,
>
>{you wrote...}
>>Sound Fonts have a field for the actual frequency.
>
>Yes, but (naturally) there aren't currently any supported, useable
>soundfont editors. The PD one's I've tried are terrible or crash, and
>original ones like Vienna require a SoundBlaster card. I'm waiting for
>a new version of Extreme Sample Convertor, due in about a month, that
>may allow for sample conversion and editing with this in mind.
>
>Still haven't heard from Prent...
>
>Cheers,
>Jon

🔗Jon Szanto <jszanto@...>

1/25/2006 6:16:23 PM

Carl,

{you wrote...}
>What about Reason, Halion, GigaStudio, Kontakt?

Don't be cryptic - what about them? The only one that I know of that currently can do any kind of internal microtuning of samples is Kontakt (if the others do, I'd be interested to know, and how implemented). And with Kontakt, you've got to write a script to do it (which I've looked into) as well as a very expensive instrument that comes with a gotcha in the manner of a c/r registration policy. I've considered Kontakt, but not until the main author of the scripting facility can get back to me on how to implement non-EDO tunings.

Share your knowledge, Mr. Lumma! :)

Cheers,
Jon

🔗Carl Lumma <ekin@...>

1/25/2006 6:20:16 PM

At 06:16 PM 1/25/2006, you wrote:
>Carl,
>
>{you wrote...}
>>What about Reason, Halion, GigaStudio, Kontakt?
>
>Don't be cryptic - what about them? The only one that I know of that
>currently can do any kind of internal microtuning of samples is
>Kontakt (if the others do, I'd be interested to know, and how
>implemented). And with Kontakt, you've got to write a script to do it
>(which I've looked into) as well as a very expensive instrument that
>comes with a gotcha in the manner of a c/r registration policy. I've
>considered Kontakt, but not until the main author of the scripting
>facility can get back to me on how to implement non-EDO tunings.
>
>Share your knowledge, Mr. Lumma! :)
>
>Cheers,
>Jon

They all open sf2 files, and then you can edit them. I don't
know if you can save the edits back to sf2 or not...

I didn't understand anything about microtunability in this:

"Yes, but (naturally) there aren't currently any supported, useable
soundfont editors. The PD one's I've tried are terrible or crash, and
original ones like Vienna require a SoundBlaster card."

-Carl

🔗Jon Szanto <jszanto@...>

1/25/2006 7:31:24 PM

Carl-dad,

:)

{you wrote...}
>They all open sf2 files, and then you can edit them. I don't know if you can save the edits back to sf2 or not...

Yes, but unless you're editing tuning, it matters not.

>I didn't understand anything about microtunability in this:

The thread was about tuning samples for microtonal use. You just didn't read far enough back, I guess. But you've reminded me to see if Bill Sethares got any further with the scripting engine in Kontakt, as he was on it as well. Bill?

Cheers,
Jon

🔗Bill Sethares <sethares@...>

1/27/2006 1:04:49 AM

--- In MakeMicroMusic@yahoogroups.com, Jon Szanto <jszanto@c...> wrote:

> The thread was about tuning samples for microtonal use. You just
didn't read far enough back, I guess. But you've reminded me to see if
Bill Sethares got any further with the scripting engine in Kontakt, as
he was on it as well. Bill?

Yes, I did get a bit further, but then I found out about LMSO:

http://www.nonoctave.com/tuning/LilMissScaleOven/instruments.html

which has a very complete tuning implementation for Kontakt. It
basically operates by writing scripts and you can read in any scale
(from Scala or a .tun file, for instance) and it automatically writes
a script containing the retuning information. You can have a large
number of different scales in the script file and can switch between them.

So, I'm very happy with this.

In fact, here is a piece in 10-tet made with a Kontakt "chamber
orchestra" (cello, flute, bassoon, recorder, clarinet) made from the
factory sounds by spectrally remapping the samples.

http://www.cae.wisc.edu/~sethares/MysteryX.mp3

The "X" stands for 10... questions or comments welcome...

--Bill

🔗Jon Szanto <jszanto@...>

1/27/2006 8:28:51 AM

Bill!

Bingo, I knew you'd come through! When researching Kontakt's capabilities, I remember you were one of the first to post and contact the guy who was responsible for the scripting code, and now with the combination of X. J. Scott's LMSO you've got an outstanding solution. Geez, am I gonna have to buy a Mac Mini just to run LMSO???

I'll check out your piece, and honest-to-goodness, I might contact Jeff to see if he would offer custom conversions to scripts for those on the PC platform.

Thanks Bill, and it's been too long since we've heard a piece of yours anyway!

Cheers,
Jon

🔗Rick McGowan <rick@...>

1/27/2006 8:41:29 AM

Jon, Bill --

> Geez, am I gonna have to buy a Mac Mini just to run LMSO???

Maybe it's not that hard to turn a .tun file into a script for Kontakt.
For example, perhaps a not-too-complicated Perl script could easily do the
trick. It partly epends on how complicated the Kontakt scripting language
is. If there were a script spec for Kontakt it might be worth looking into,
for people who own Kontakt.

Rick

🔗Bill Sethares <sethares@...>

1/27/2006 9:31:22 AM

Jon Szanto wrote:

> Bingo, I knew you'd come through! When researching Kontakt's
capabilities, I remember you were one of the first to post and contact
the guy who was responsible for the scripting code, and now with the
combination of X. J. Scott's LMSO you've got an outstanding solution.
Geez, am I gonna have to buy a Mac Mini just to run LMSO???
>
> I'll check out your piece, and honest-to-goodness, I might contact
Jeff to see if he would offer custom conversions to scripts for those
on the PC platform.

Jeff is truly a "man with a mission" -- he's dedicated and thorough.
When I found a tiny bug in the software, he emailed me a revised in
about 2 days!

I had looked into the Kontakt scripting the language, and even written
a small script that does equal temperaments... it was obvious that
there must be some way to do it for general tunings (maybe Rick's
suggestion of a scipting language be a way to do it) but it was also
obviously an incredible amount of work. I was very "psyched" when I
found out that Jeff had already done it. LMSO is not cheap, but heck,
neither is Kontakt!
>
> Thanks Bill, and it's been too long since we've heard a piece of
yours anyway!

Enjoy...

--Bill

🔗Jon Szanto <jszanto@...>

1/27/2006 9:02:30 AM

Rick,

{you wrote...}
>Maybe it's not that hard to turn a .tun file into a script for Kontakt.

Well, that is what I'm thinking at this point. I know enough about Perl and the derivatives that I could probably do it. I've looked at some of the microtuning scripts they have, but I've only seen EDO ones; if I could look at one or two non-EDO scripts I'd probably have all the logic I need to do the conversions.

Of course, I'd have to buy Kontakt as well! Dare I give money to NI? :)

Cheers,
Jon

🔗Carl Lumma <ekin@...>

1/27/2006 12:19:25 PM

At 08:28 AM 1/27/2006, you wrote:
>Bill!
>
>Bingo, I knew you'd come through! When researching Kontakt's
>capabilities, I remember you were one of the first to post and contact
>the guy who was responsible for the scripting code, and now with the
>combination of X. J. Scott's LMSO you've got an outstanding solution.
>Geez, am I gonna have to buy a Mac Mini just to run LMSO???

Perhaps you can run it under OS X on your PC... that's what I
hope to do.

-Carl

🔗Jon Szanto <jszanto@...>

1/27/2006 12:58:03 PM

Carl,

{you wrote...}
>Perhaps you can run it under OS X on your PC... that's what I hope to do.

I am totally unaware of this ability (I've heard of SoftWindows or something on the Mac). Drop me a line off-list to bring me up to speed, sounds promising!

Cheers,
Jon