issue13

EMUSIC-L Digest                                      Volume 51, Issue 13

This issue's topics:
	
	Sound Morphing (14 messages)

Your EMUSIC-L Digest moderator is Joe McMahon .
You may subscribe to EMUSIC-L by sending mail to listserv@american.edu with 
the line "SUB EMUSIC-L your name" as the text.
 
The EMUSIC-L archive is a service of SunSite (sunsite.unc.edu) at the 
University of North Carolina.
------------------------------------------------------------------------
Date:         Tue, 20 Apr 1993 21:06:00 EDT
From:         rg8 
Subject:      


>>As has been mentioned, I don't see how you can do this with MIDI
>>instruments, since the general architecture has dead ends -- or at least
>>limits -- for transforming one patch to another.
>
>Well, existing instruments themselves I'm sure can't do it. However, a
>decent sampler with a SCSI connection ought to be able to hand me a file or
>files I  can mess with and load back again. This is just speculation,
>really, but if
>it stimulates someone to experiment, it was worthwhile. It's even more
>worthwhile if I get to hear it sometime.

I agree, and this is why I'm going to work with Csound as soon as I get a
fast computer.  Studio Vision also comes to mind (I don't plan to give up all
my MIDI gear).

BTW, there have been some notable pieces for orchestra which explore the idea
of transformation.  My favorites are Gyorgy Ligeti's pieces for large
orchestra:  Melodien, San Franciso Polyphony and, especially, Lontano.

Bob Gibson
rg8@umail.umd.edu

------------------------------
Date:         Wed, 21 Apr 1993 10:02:05 GMT
From:         Martin Rootes 
Subject:      Re: Transmogrification

> Literally, a transformation of a sound from the standpoint of something
> like the T2 transformation from a tile floor through flowing metal into a
> human form.  I want a sound to start as a flute, shift into filtered pink
> noise, and then transform into a bass drum (as an example, dunno what I'd
> be able to use that particular sound for).
I think what people are looking for here is the Phase Vocoder software, this
is, I think, available as part of either csound or cmusic (or maybe both!).
The Composers Desktop Project in York (UK) do a suite of software and hardware
for STs which includes the Phase Vocoder. This stuff will not operate in real
time though, so once you've created a sound (apparantly very slow on an ST)
you'd have to transfer it to a sampler (or alternatly create you complete piece
of music by csound/CDP).

   Martin.

------------------------------------------------------------------------------
Martin Rootes - Senior Systems Programmer/Analyst, Sheffield Hallam University
Email :         M.Rootes@shu.ac.uk
------------------------------------------------------------------------------

------------------------------
Date:         Thu, 22 Apr 1993 07:53:13 GMT
From:         w.purvis@DARESBURYAC.UK
Subject:      Re: Transmogrification

The recent discussions on this topic have drawn comparisons with
transformations
of visual images. The major difficulty with such transformations is determining
which `features' of the original image correspond to which in the final image.
For instance in transforming a face its not too bad, features such as the
corners of the eyes and mouth are obvious, points along the hairline are a bit
more tricky, but then that's where the `artistic' side comes in. With music
it can be difficult to define what the significant features of a timbre are
and even when you have managed to do this, you then have to relate the features
of
one to the features of another. You could take a simplistic view that all
sounds
are just a collection of sine waves and `slide' from one spectrum to another,
but even then it's not obvious how you do it.

Let me give a simple example: Imagine a sound that consists of two sine waves
with amplitudes 100 and 200 (arbitrary units) and frequencies of 440 and 660
Hz.
Suppose we want to transform this sound to one with amplitudes 160 and 80 at
frequencies of 220 and 330 Hz. We obviously have two significant features in
both cases, but do we slide the 440 Hz to the 220 Hz or the 330Hz? The
frequency
ratios suggest that maybe it should be 220, but on the other hand we might say
that the amplitude is more important and so the 330 might be better. OK, you
can
do it two ways, but that's only looking at it in one way. We could say that
both
sounds have components at all four frequencies (two with zero amplitude in both
cases) and just fade one pair down while the other fades up - this is the
conventional cross-fade approach. They are all valid approaches for getting
from A to B, they just go by different routes. The idea that you can `program'
such a transformation on more complex sounds seems pretty far-fetched to me.
You need to supply an awful lot of supplementary information to the process.
I guess this is where the art comes in!

Bill.

------------------------------
Date:         Mon, 26 Apr 1993 02:43:04 GMT
From:         Scott Fisher 
Subject:      Re: transmogrify

Joe McMahon  writes:

>Yes. Certainly one has a lot of options here: assign "close" points (i.e.,
>choose peaks and valleys in the two waveforms that would be close to one
>another if the waveforms were overlaid) or "distant" points (randomly
>select places in the waveforms to be mapped into one another).

I am not sure anyone has mentioned this...

I saw somme Atari software that let you "morph" from one sample to another.
I think it only worked with short (several cycles) but I can't be sure.
You show the computer the start and end waveforms and it morphs between the
two at the rate you specify. The example I saw was going from a triangle
wave to a piano wave (I think).

Sorry I am short on details, I saw the software reviewed in either
Music-Technology or Sonics.

I'll try to re-locate the article.

Regards Scott.
_______________________________________________________________________________
Scott Fisher [scott@psy.uwa.oz.au]  PH: Aus [61] Perth (09) Local (380 3272).
                                                             _--_|\       N
Department of Psychology                                    /      \    W + E
University of Western Australia.      Perth [32S, 116E]-->  *_.--._/      S
Nedlands, 6009.  PERTH, W.A.                                      v

    *** ERROR 144 - REBOOT? is a registered trademark of ENSONIQ Corp ***
-------------------------------------------------------------------------------

------------------------------
Date:         Thu, 29 Apr 1993 14:30:55 -0400
From:         Joe McMahon 
Subject:      Re: Morphing from One Waveform to Another

>For all of you morphing fans and ponderers:
>
[complicated but fascinating process deleted]

>  What do you guys think of the process (besides its general kludginess)?
>
Sounds great! Let us know when you've actually gotten to try it.

 --- Joe M.

------------------------------
Date:         Thu, 29 Apr 1993 14:16:52 -0500
From:         Brian Adamson 
Subject:      Morphing from One Waveform to Another

For all of you morphing fans and ponderers:

  I've figured out the pieces to try out this Morphing idea and actually
have most of the pieces.  Here are the steps I propose just to see what it
sounds like using a Mac, some software,and an EPS 16+:

1)  Create a text data table of the samples for two different single cycle
waveforms. (start and end point)

2)  Open each file with Cricket graph, plot it, and save the files as PICT
files on the Mac.

3)  Open the PICT files with some shareware morphing software I've just
downloaded for the Mac (called Metamorphisis).

4)   Create a QuickTime movie using the Metamorphisis software of N (e.g.
15) frames, morphing between the start and end point sounds (I guess I'll
just take a seat-of-the-pants approach to correlating points for the morph
process on the two single cycle waveforms)

5)  Dissect the resulting QuickTime movie into individual frames (PICT
files) using cut & paste on the Mac.

6)   Using one of these software packages I've seen ads for which can
convert a pictoral graph of something back into a table of numbers, convert
the PICT files into Raw sample files (this is the one piece of software I
don't have yet)

7)   Convert the raw sample files into AIFF files using some software I've
hacked together for past experiments.

8)   Using Sound Designer II on the Mac, convert the AIFF files together,
to Sound Designer II format.

9)  Using Terje Finstad's EPSm Mac software, import the AIFF files into EPS
samples on floppy and load them into the EPS (This is faster than
transferring the sounds to the EPS via MIDI)

10)  I can then assemble the single cycle waveforms, and use the EPS 16+'s
TRANSWAVE feature to slide through (i.e. morph) the wavetables from one
waveform to another ...

11)  Listen to result.


  I may try this when I complete my move to a new house and get back from
the UK (end of May).

  What do you guys think of the process (besides its general kludginess)?


- Brian A.


 ____________________________________________________________________
|                     _ _ _                                          |
|        /|\    /\  /\_ _ /\  /\       R. Brian Adamson              |
|       / ||   / / / /   / / / /      Naval Research Laboratory      |
|      / /||  / / /_/_ _/ / / /      Information Technology Division |
|     / / || / / / \_|\_\/ / /      Code 5523                        |
|    / /  ||/ / / /  ||   / /                                        |
|   / /   |/ / / /   ||  /_/_ _   adamson@itd.nrl.navy.mil           |
|   \/     \/  \/     \  \ _ _ _\                                    |
|____________________________________________________________________|

------------------------------
Date:         Fri, 30 Apr 1993 08:34:34 -0400
From:         idealord 
Subject:      Re: Morphing from One Waveform to Another

Sounds totally ridiculous!  Not to be rude - but there are real sound data
morphing processes - probably the most interesting being linear predictive
resynthesis.  Admittedly, it's a slow process, but certainly no slower than
reducing waves to graphic information and then morphing the graphs.

You are bound to get all kinds of _bad_ acoustic artifacts from the graphical
processes - because - you're not mapping acoustical parameters - you're
mapping maps or graphics over and over.

Linear predictive synthesis can be used like filter convolving - basically
you're creating a finite set of filters which when activated by a leftover bit
of noise reynthesizes the waveform.  When you cross-fade the filter maps you
get very interesting sound permutations - this is how the guys at Stanford do
the old talking orchestra trick - this is what Paul Lansky uses to get his
bizarre transformations.

In F. Richard Moore's book - _Elements of Computer Music_ there's C code for
doing Linear Predictive synthesis - I've always wanted to fool around with it
but on my Amiga 1000 it would take 12 hours to do one filter map.  For those
with powerful Macs - you should check out CSound - it will let you do LP
processes now.

Also - check out SoundHack - Tom Erbe (sp?) great little Mac tool for doing
file transformations - it lets you do convolving - you should be able to
cross-fade a couple of filter maps and get some totally cool transformations
on your samples!  It's at mills.edu - check it out...

Jeff Harrington
idealord@dorsai.dorsai.org

------------------------------
Date:         Fri, 30 Apr 1993 09:23:49 -0400
From:         Joe McMahon 
Subject:      Re: Morphing from One Waveform to Another

>Sounds totally ridiculous!  Not to be rude - but there are real sound data
>morphing processes - probably the most interesting being linear predictive
>resynthesis.  Admittedly, it's a slow process, but certainly no slower than
>reducing waves to graphic information and then morphing the graphs.
>
>You are bound to get all kinds of _bad_ acoustic artifacts from the graphical
>processes - because - you're not mapping acoustical parameters - you're
>mapping maps or graphics over and over.

True, but sometimes artifacts are *good*. I won't pass judgement on this
method until it's been tried. It's certainly the closest thing I've heard
to research in a while. What the heck, if it doesn't work, it doesn't. But
it will be fun to try.

 --- Joe "Serendipity? Yeah." M.

------------------------------
Date:         Fri, 30 Apr 1993 09:37:54 -0400
From:         idealord 
Subject:      Re: Morphing from One Waveform to Another

>
> >Sounds totally ridiculous!  Not to be rude - but there are real sound data
> >morphing processes - probably the most interesting being linear predictive
> >resynthesis.  Admittedly, it's a slow process, but certainly no slower than
> >reducing waves to graphic information and then morphing the graphs.
> >
> >You are bound to get all kinds of _bad_ acoustic artifacts from the graphical
> >processes - because - you're not mapping acoustical parameters - you're
> >mapping maps or graphics over and over.
>
> True, but sometimes artifacts are *good*. I won't pass judgement on this
> method until it's been tried. It's certainly the closest thing I've heard
> to research in a while. What the heck, if it doesn't work, it doesn't. But
> it will be fun to try.
>
>  --- Joe "Serendipity? Yeah." M.
>

Well......one of the things which makes computers interesting is - the
infinite things there are to _try_.  Those with the luxury of time can ignore
other people's research - I'm not knocking creative mucking about, now - but
if what you want to do is really morph sounds - there are real ways to do it,
ways which have not been exploited one iota beyond the academic... I would
love to see everyone fooling around with resynthesis - which is what we're
talking about - but he asked for comment and I commented - it's a naive idea -
a dead end...

As you may know - there are also an infinite number of ways to create
artifacts - if you want to morph sounds well - find a _real_ process and be
creative - you're going to get artifacts every single time you subject your
sounds to a process - I thought the point was to get cool, creative, new
effects.  Whatevever... it's your time!  (Laughing all the way to the time
bank!)

Jeff Harrington
idealord@dorsai.dorsai.org

------------------------------
Date:         Fri, 30 Apr 1993 10:25:55 EDT
From:         ronin 
Subject:      chill, bud

criticism of technique is always welcome.
half-witted, ill-considered rants are worthless, regardless of the merit of
whatever 'real' information may be contained therein.
btw, idealord seems to have missed the point of analysis/resynthesis.
there is direct relation between waveform modification and
spectral content. to suggest that artifacts may occur in a process
operating on one domain *because it bears no relation to the other
domain* is obtuse.
it is interesting how people who advertise themselves as intelligent,
open thinkers sometimes turn out to be the most conservative, opposed
in principle to the simple process of intellectual game-playing.

-----------< Cognitive Dissonance is a 20th Century Art Form >-----------
Eric Harnden (Ronin)
 or 
The American University Physics Dept.
4400 Mass. Ave. NW, Washington, DC, 20016-8058
(202) 885-2748  (with Voice Mail)
---------------------< Join the Cognitive Dissidents >-------------------

------------------------------
Date:         Fri, 30 Apr 1993 10:58:42 EDT
From:         Brian Good 
Subject:      Re: Morphing from One Waveform to Another

( lots of yelling about sound morphing deleted )

I haven't seen this discussed here, so excuse me if this is commmon knowledge,
but--for quick and dirty experimentation with sound, Mathematica 2.0 and
higher (on some platforms) has Play and ListPlay objects that can play either
functions or lists of samples.  These tend to be implemented on platforms
like the Mac that aren't notoriously fast, so cpu-intensive things like morphing
might not work well, but you can still do quite a lot.  Mathematica has a pretty
good-sized list of built-in functions, which can save you a lot of code-writing
time (e.g. it knows about Bessel functions so you can do simple FM more or less
instantly).

brian good
bgood@sundagger.lerc.nasa.gov

------------------------------
Date:         Fri, 30 Apr 1993 10:15:55 -0500
From:         Brian Adamson 
Subject:      Re: Morphing from One Waveform to Another

>>
>> >Sounds totally ridiculous!  Not to be rude - but there are real sound data
>> >morphing processes - probably the most interesting being linear predictive
>> >resynthesis.  Admittedly, it's a slow process, but certainly no slower than
>> >reducing waves to graphic information and then morphing the graphs.
>> >
>> >You are bound to get all kinds of _bad_ acoustic artifacts from the
>>graphical
>> >processes - because - you're not mapping acoustical parameters - you're
>> >mapping maps or graphics over and over.
>>
>> True, but sometimes artifacts are *good*. I won't pass judgement on this
>> method until it's been tried. It's certainly the closest thing I've heard
>> to research in a while. What the heck, if it doesn't work, it doesn't. But
>> it will be fun to try.
>>
>>  --- Joe "Serendipity? Yeah." M.
>>
>
>Well......one of the things which makes computers interesting is - the
>infinite things there are to _try_.  Those with the luxury of time can ignore
>other people's research - I'm not knocking creative mucking about, now - but
>if what you want to do is really morph sounds - there are real ways to do it,
>ways which have not been exploited one iota beyond the academic... I would
>love to see everyone fooling around with resynthesis - which is what we're
>talking about - but he asked for comment and I commented - it's a naive idea -
>a dead end...

   I did a little creative mucking about ... not so much as a result of
having luxury of time but more so of coming into some "mucking about" tools
... Anyway, I put together a little QuickTime movie of morphing a triangle
single cycle to a square ... Agreed, the result has little practical value
... All experiments are *not* necessarily successful.  Here are some
comments:

The result is kind of cute but not very practical to turn into a set of
"transwaves" (at least not without a *lot* of interpretation)

   However, I was reminded by performing this task of some certain
fundamental 9th grade mathematical rules which govern such functions as
single-cycle waveforms.  That is, that they are functions with respect to
time (at least mono sound) in one dimension only ... amplitude (sound
pressure) versus time ... Graphic 2D images are 2 dimensional and of course
don't obey the rules of a graphical representation of a linear function
(i.e. there is one distinct y point for each corresponding x-axis (time)
point).

   This simple rule governing one-dimensional, linear functions, really
only allows freedom of movement in one dimension ...the y-axis ... which
really ends up being just some variant of a crossfade I think (note that
this is in the time domain only, maybe something differently could be done
in the frequency domain)... I think we've all been morphing audio for long
time without thinking of it as such.  It just isn't as psychologically
dramatic in the relatively 1D mono-audio world as it is in the graphics
world.   Note that perhaps something interesting might be done with stereo
sound that leaves the same type of dramatic impression that 2D image
morphing does. Hmm....

    Anyway this little QuickTime movie illustrates this a little ...(I
mailed in BinHex form to Joe to post on Castrovalva for interested parties.

>
>As you may know - there are also an infinite number of ways to create
>artifacts - if you want to morph sounds well - find a _real_ process and be
>creative - you're going to get artifacts every single time you subject your
>sounds to a process - I thought the point was to get cool, creative, new
>effects.  Whatevever... it's your time!  (Laughing all the way to the time
>bank!)

   I agree that LPC offers some possibilities here as a real_process for
morphing among other things. The vocoding-like application that SoundHack
performs is certaintly interesting ... It's probably worth mucking about a
little, too ... Does the SoundHack algorithm go very far in extracting
excitation characteristics in addition to the filtering properties?  The
copy I have only does Phase Vocoding, convolution, some others.  Disabled
menu items include a "Mutation" process ...I wonder what the author has in
mind here?

>
>Jeff Harrington
>idealord@dorsai.dorsai.org
 ____________________________________________________________________
|                     _ _ _                                          |
|        /|\    /\  /\_ _ /\  /\       R. Brian Adamson              |
|       / ||   / / / /   / / / /      Naval Research Laboratory      |
|      / /||  / / /_/_ _/ / / /      Information Technology Division |
|     / / || / / / \_|\_\/ / /      Code 5523                        |
|    / /  ||/ / / /  ||   / /                                        |
|   / /   |/ / / /   ||  /_/_ _   adamson@itd.nrl.navy.mil           |
|   \/     \/  \/     \  \ _ _ _\                                    |
|____________________________________________________________________|

------------------------------
Date:         Fri, 30 Apr 1993 12:03:35 -0400
From:         idealord 
Subject:      Re: Morphing from One Waveform to Another

>    I agree that LPC offers some possibilities here as a real_process for
> morphing among other things. The vocoding-like application that SoundHack
> performs is certaintly interesting ... It's probably worth mucking about a
> little, too ... Does the SoundHack algorithm go very far in extracting
> excitation characteristics in addition to the filtering properties?  The
> copy I have only does Phase Vocoding, convolution, some others.  Disabled
> menu items include a "Mutation" process ...I wonder what the author has in
> mind here?
>
> >
> >Jeff Harrington
> >idealord@dorsai.dorsai.org

I've never seen the thing - friend of mine turned me on to it - hoping it
would do LPC - but it doesn't...  Again - why doesn't someone type in the C
code in "Elements of Computer Music" and we can all give LPC a shot.  As soon
as you've got the filter elements and cross-fade them with another group of
filter elements - any transformation is possible.  Talking pianos, electric
guitars that turn into Tibetan horns.... You'd better have a computer with a
fast FPU!

Oh, yeah, quick comment to ronin?(sp) - no comment! ;-)  We don't need a flame
fest here on a mailing list - my comments about the proposed graphical process
were intended as comment (as requested!).  Judge not or thyself be judged!
Yawn....

Jeff Harrington
idealord@dorsai.dorsai.org

------------------------------
Date:         Fri, 30 Apr 1993 17:37:53 -0700
From:         Jesse Kanner 
Subject:      Re: Morphing from One Waveform to Another

Since you know about Richard Moore and C-Sound, I would assume that
you are familiar with the CARL facility at UCSD. I studied electronic
music there with Marc Battier (from IRCAM) and Gordon Mumma (from
Mars..I think...). C-Sound  and many of the other resouces developed at
UCSD are totally bitchin. I too would urge all to serious check out C-Sound.

Open questions: Has anybody heard about David Tudor? Is he still producing
work? I find him to be one of the Chosen Few in the electronic music world.
Also, there was a grad, student at UCSD named Eric Lyon. Does anyone know
his E-Mail address? And if so, how could they let him even touch a computer?

------------------------------
End of the EMUSIC-L Digest
******************************