Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!apple!motcsd!lance
From: lance@motcsd.csd.mot.com (lance.norskog)
Newsgroups: comp.compression
Subject: Re: Sound compression
Keywords: sound compression
Message-ID: <3958@motcsd.csd.mot.com>
Date: 15 May 91 19:54:14 GMT
References: <16198@helios.TAMU.EDU> <19524@crdgw1.crd.ge.com> <104045@sgi.sgi.com> <91May15.140250edt.750@neuron.ai.toronto.edu>
Organization: Motorola CSD, Cupertino CA
Lines: 14

Back to sound compression...

Examination of the data always helps.  If you look at a sound sample
(I use MixView, a marvelous package from two guys at Columbia, contact
doug@woof.columbia.edu) you will notice that it's usually a regular
waveform.  I have this theory that you can do compression via 
delta modulation by seeking forward X samples and hunting for a
nearby sample.  With regular waveforms, X will tend to be constant
for interesting periods of time.

I like the 14->8 trick very much.  If I didn't have a stinking Sound
Blaster I'd try it.

Lance