DEFINITION OF WORD

[Curtin, Steven D (Steven)]

Dave,

And what a word it is!  Convolution is very well explained in both the F.R.
Moore and Charles Dodge Computer Music books.  The basic idea is that you
take a delay line and turn it into a filter by taking each sample and
multiplying it with a corresponding sample in another delay line, like this:

for N:
output sample += input sample[n] * convolution sample[n];

And yes indeed you can use convolution to "filter" one sound with another.
I put together an app in the early 90's the downloaded the convolver table
from a hand-drawn table to a DSP, drawing a sharp wave would let many
frequencies through, drawing a smooth wave made the convolver loop output
sound like a low pass filter.  This technique is used a lot for HRTF
modelling for 3D sound, etc.  In Cmix which is used at Princeton, you can in
non-real-time convolve one sound with an entire sound file and the results
are amazing.

This is primarily a "digital" technique, but there's no reason you couldn't
do this in analog with a couple of bucket brigades, some counters,  a TCA or
other multiplier, and a mixer with s/h to add up all the samples.  I'm just
thinking out loud here.  The bucket brigade would have to have addressable
access to all the samples.

Steve C

-----------------------------------------------------------------
Steven Curtin  
Lucent Technologies Microelectronics
ph: (732)949-4404   fax: (732)949-6711
http://curtin.emf.org
sdcurtin at lucent.com
-----------------------------------------------------------------


> ----------
> From: 	dsolursh at georgian.net[SMTP:dsolursh at georgian.net]
> Sent: 	Monday, March 06, 2000 12:24 AM
> To: 	synth-diy at mailhost.bpa.nl
> Subject: 	DEFINITION OF WORD
> 
> Hay there, I have come across a word and technique that I am not totaly
> sure about and would like to understand in detail.  The word is
> "convolution" and seems to apply to a technique or manipulation that
> resembles frequency modualtion, but seems to filter the sound with
> another's envelope.  Could anyone help me to find some info or to
> understand what this technique involves.  Thanks, Dave.
> 
>