Use hetro (analysis utility) and adsyn (synthesis opcode) for time/frequency
analysis and resynthesis using a filterback analysis and oscillator bank
resynthesis.

You can read the analysis files with Java, monkey with them, and write them
back again. You can even translate other data (pictures, for example) into
analysis files and synthesize them.

My software Silence (currently in interminable beta release) does all of
this. http://ww.pipeline.com/~gogins.

-----Original Message-----
From: Andreas Schoter 
To: csound@maths.ex.ac.uk 
Date: Wednesday, December 23, 1998 10:50 AM
Subject: Newbie Intro & Question


Hi all,

As per the joining recommendation, here's a brief bit about me, and my
first question.

I've just started running CSound under Win95 - I prefer Linux for most
things, but all my other music software (VAZ softsythn, MIDI sequencer etc)
is on windows so it seemed silly to have CSound as the only Linux app -
especially when I haven't really configured my sound card under Linux, and
PnP is a pain with my current kernel :-)

I have no formal musical training, but I have some previous experience with
analogue sythns (I owned an EDP Wasp when they first came out).  So, I'm
just winging it here.

My first composition project with CSound is going to be based on additive
synthesis, because I have some understanding of that.  The idea is that I'm
going to set up a set of sine wave ocillators at different frequencies
using the oscil opcode and vary the amplitude of each over time to give me
a subtly shifting series of tones.  I'm sure this has been done before, but
there you go.

OK - so here's my question.  The obvious way to do this is with a separate
instrument for each voice, with each instrument having its own linseg
controlling the amplitude of the wave over time.  This is the way I'm
working at the moment.

But I'm open to suggestions as to a more generic way of accomplishing this.

Also, to help me with these explorations, I'm planning to build a Java
front end targetted at this kind of additive synthesis.  My plan is to have
a window for each channel, with a selectable number of harmonics in each
window.  Each harmonic would have a selectable waveform (limited to sine in
the first instance).  The time/amplitude function for each voice could then
be drawn into the window, with each harmonic being displayed in a distinct
colour so that you can get a good overview of the whole channel at once.
Once the compoisition is complete the program would then generate the
appropriate orc and sco files.

If there is already something like this, I wouldn't mind being spared the
effort :-)  Otherwise, I'll let folks know how it goes.

Many Thanks

Andreas

________________________________________________________________________
                                     Dr Andreas Schöter, Intertrader Ltd
                                              http://www.intertrader.com
                        Tel: +44(0)131 475 7108, Fax: +44(0)131 475 7109