Yahoo Tuning Groups Ultimate Backup makemicromusic Prent: sample question

Hey Mr. Prent Rodgers!

(I'm asking in public to allow this knowledge wider range)

When you've worked over the years in preparing samples for your use with Csound (I believe you were using some of the McGill sample set), have you taken and resampled individual notes to fit a particular tuning scheme (frequently the Partch diamond), or do you stick with the 12tet sample set provided and allow the software to retune, as needed, the closest 12tet pitch to the actual pitch you desire?

I'm trying to map out a strategy for sample usage in this coming year, and your advice would be helpful.

Cheers,
Jon

P.S. Work must be thick - been a long time since a new PR piece!

🔗Prent Rodgers <prentrodgers@...>

Jon,
Csound can take a sample of an instrument playing any pitch, and
generate any other pitch. You have to tell it what pitch the sample
has recorded, and what pitch you want to hear, and it does the
calculation to play the note at the pitch you want. It's just mathmatics.

For example, the following line of Csound code:

a3,a4 loscil 1, kcps2, ifno, ibascps

a3 and a4 are the output waveforms, left and right channel
loscil is the "opcode" that does the work
1 sets the output level to a known volume (I later change it by
applying an envelope)
kcps2 is the frequency that I want the note to play
ifno is the number of a storage location that contains the sample in
memory
ibascps is the frequency recorded in the sample

There is other code around that code to determine the right sample to
use, to calculate an offset in case the recorded note is out of tune
(they all are), and the frequency to choose based on the tuning system.

I've been busy creating a new set of samples based on the University
of Iowa sample library, starting with a brass quintet. I'm about three
months into it, and hope to get another piece out in another three months.

Prent Rodgers

--- In MakeMicroMusic@yahoogroups.com, Jon Szanto <jszanto@c...> wrote:
>
> Hey Mr. Prent Rodgers!
>
> (I'm asking in public to allow this knowledge wider range)
>
> When you've worked over the years in preparing samples for your use
with Csound (I believe you were using some of the McGill sample set),
have you taken and resampled individual notes to fit a particular
tuning scheme (frequently the Partch diamond), or do you stick with

🔗Jon Szanto <jszanto@...>

Prent,

Thanks for the info. I probably should have made my question even more straight-forward, but seeing your answer I can frame it this way: "Does Prent utilize 12tet samples originally, and then recalculate the nearest sample to the desired pitch?" and the answer would seem to be "Yes".

It is looking like there might be at least two routes to getting this done for me as well, but one thing I'm looking at as a side benefit is having one of the instruments/applications apply a tuning to a set of samples, and once I am happy with the results, using an automated tool to generate a new set of samples in the new tuning. There are a couple of VST hosts that will generate an entire set of pitchs, at whatever velocities you like, into new waveforms, which one could then export to any sampler that didn't even utilize tuning schemes.

{you wrote...}
>It's just mathmatics.

Hey, you can't scare me off that easily. :)

>I've been busy creating a new set of samples based on the University of Iowa sample library, starting with a brass quintet. I'm about three months into it, and hope to get another piece out in another three months.

Six months? What a slacker. Next you'll try to convince me all the rain has slowed you down... :)

Cheers,
Jon

🔗Prent Rodgers <prentrodgers@...>

--- In MakeMicroMusic@yahoogroups.com, Jon Szanto <jszanto@c...> wrote:
>
> Prent,
>
> Thanks for the info. I probably should have made my question even
more straight-forward, but seeing your answer I can frame it this way:
"Does Prent utilize 12tet samples originally, and then recalculate the
nearest sample to the desired pitch?" and the answer would seem to be
"Yes".
>

I tell a preprocessor which of the 37+ notes/octave I want Csound to
play, and it determines which samples to include. Then at sound
generating time, Csound takes the note number I pass it, and does a
table look up to determine which sample to use. It also finds out how
many cents to adjust the sample so that it is in tune with 12 TET, and
what the base frequency it should use. It then does a table look up to
determine the desired output frequency, and generates the note. I also
apply a multiplier to the frequency for things like vibrato and
glissando. And an envelope. It turns out the the vibrato, glissando,
and envelope are far more important to realistic sounding samples than
the original samples. If you apply a guitar envelope to an oboe, it
sounds like a guitar.

Prent

🔗Gene Ward Smith <gwsmith@...>

🔗Prent Rodgers <prentrodgers@...>

--- In MakeMicroMusic@yahoogroups.com, "Gene Ward Smith"
<gwsmith@s...> wrote:
>
> --- In MakeMicroMusic@yahoogroups.com, "Prent Rodgers"
> <prentrodgers@c...> wrote:
>
> It turns out the the vibrato, glissando,
> > and envelope are far more important to realistic sounding samples than
> > the original samples. If you apply a guitar envelope to an oboe, it
> > sounds like a guitar.
>
> Wouldn't this apply to purely synthesized Csound instruments also?
>

Yes it does. The thing that makes an instrument interesting is the way
it changes for every note. If you are careful about applying the right
physical models to a synthesized sound, you can make it as interesting
as a real instrument. But if every note sounds the same, the
synthesized instrument quickly sounds synthesized. Play one note and
you sometimes can't tell. But if you play several in a row, and they
all sound the same, anyone can tell they are not real. It's the
changes that make them interesting.

Prent

🔗Jon Szanto <jszanto@...>

🔗Gene Ward Smith <gwsmith@...>

🔗Prent Rodgers <prentrodgers@...>

--- In MakeMicroMusic@yahoogroups.com, "Gene Ward Smith"
<gwsmith@s...> wrote:

>
> So where would one find information on what would be good values to
> apply? To some extent formats do this, but I wonder if that is still
> to mechanical.
>

Well, I would say there are no "good values", only ones that make
sense at that moment of the composition. Google defines it:

"Physical modelling synthesis is the synthesis of sound by using a set
of equations and algorithms to simulate a physical source of sound.
Sound is then generated using parameters that describe the physical
materials used in the instrument and the user's interaction with it,
for example, by plucking a string, or covering toneholes, and so on."

It is the abundance of parameters that make it so powerful. You have
so much to play with.

I've enjoyed what the physical modeling synthesizer composers have
produced. See Julius Smith's material at
http://ccrma.stanford.edu/~jos/pmupd/pmupd.html . Csound has some
implementations of physical modeling synthesizer building blocks. Eric
Kuehnl has a basic intro to "pluck", the first of many physical model
opcodes for csound here: http://music.calarts.edu/~eric/phys.html .

The Csound Book (http://www.powells.com/biblio?isbn=0262522616) has a
chapter on physical modeling.

🔗Carl Lumma <ekin@...>

Perhaps some confusion was due to Prent's original language, which
spoke of applying physical models TO a synth sound. There ARE
physically modeled effects (like reverb), but here Prent is talking
about pure PM synthesis, where the model is the origin of the
sound.

There are now several commercial products using this approach. The
Yamaha VL series hardware synths, discussed previously on this list,
have been around since the mid '90s (and represent sort of an attempt
to have another FM-like success by commercializing something from
Stanford, though it doesn't seem to have worked). The Korg OASYS PCI
sound card, from the late '90s (no longer produced, but maybe on
ebay), had a fantastic PM electric piano (and maybe others). I think
this technology (and maybe even the patches themselves) made it in to
the new OASYS super-high-end keyboard from Korg. While few of us can
afford an OASYS keyboard, Korg has promised trickle-down products,
the first of which is the RADIAS

http://korg.com/gear/info.asp?a_prod_no=RDKB&category_id=1

though it's not clear what aspect of this is PM.

AAS has Lounge Lizard (electric piano) and String Studio (amazing)
are very good products.

http://www.applied-acoustics.com/stringstudio.htm

String Studio and Cameleon 5000 are probably my two favorite synths
ever.

There is also "circuit modeling", which is a kind of PM synthesis,
widely used for recreating analog synths of the past (Moog, etc.).

"Formats" -- you meant "formants", Gene? They're a big part of
the RADIAS synth, though I'm not sure PM is involved in that part.
One of Wendy Carlos' trademark techniques is to "shape" her
keyboard-performed synth parts with a vocoder. But the kind of
flexibility given by PM in something like String Studio, where
(say) MIDI velocity can be used to control the amount of energy
driven into the string/resonator (and the materials of the string
and resonator can be specified in more or less real time) is
simply another level beyond that.

-Carl

>> So where would one find information on what would be good values to
>> apply? To some extent formats do this, but I wonder if that is still
>> to mechanical.
>
>Well, I would say there are no "good values", only ones that make
>sense at that moment of the composition. Google defines it:
>
>"Physical modelling synthesis is the synthesis of sound by using a set
>of equations and algorithms to simulate a physical source of sound.
>Sound is then generated using parameters that describe the physical
>materials used in the instrument and the user's interaction with it,
>for example, by plucking a string, or covering toneholes, and so on."
>
>It is the abundance of parameters that make it so powerful. You have
>so much to play with.
>
>I've enjoyed what the physical modeling synthesizer composers have
>produced. See Julius Smith's material at
>http://ccrma.stanford.edu/~jos/pmupd/pmupd.html . Csound has some
>implementations of physical modeling synthesizer building blocks. Eric
>Kuehnl has a basic intro to "pluck", the first of many physical model
>opcodes for csound here: http://music.calarts.edu/~eric/phys.html .
>
>The Csound Book (http://www.powells.com/biblio?isbn=0262522616) has a
>chapter on physical modeling.

🔗Carl Lumma <ekin@...>

🔗Gene Ward Smith <gwsmith@...>

🔗Carl Lumma <ekin@...>

At 04:07 PM 1/29/2006, you wrote:
>--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@l...> wrote:
>
>> AAS has Lounge Lizard (electric piano) and String Studio (amazing)
>> are very good products.
>>
>> http://www.applied-acoustics.com/stringstudio.htm
>>
>> String Studio and Cameleon 5000 are probably my two favorite synths
>> ever.
>
>I downloaded the String Studio manual, and as usual can't figure out
>anything from it of the slightest use to me. I need to know what it
>does as far as tuning goes, and I need to know if it can convert midi
>or some other score format into music. Why does it need a sound card??

The primary use of these things is as live performance instruments.
So they usually expect you to have a sound card. You may not need
one, though. You should be able to attach it to a MIDI track in
SONAR, then 'freeze' the result to an audio file. But you should
find that live output is extremely handy in tweaking patches,
levels, etc., before doing so (I'm guessing you currently render,
listen, tweak, repeat... = time consuming).

I don't know if it does MIDI tuning, but I assume it respects
pitch bends.

-Carl