Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!uknet!pipex!uunet!munnari.oz.au!hp9000.csc.cuhk.hk!sparc15.cs.cuhk.hk!agmiller
From: agmiller@ee.cuhk.hk (Miller_Alan_George)
Subject: Software required for SPARC 
Message-ID: <CG8BB7.7x2@sparc15.cs.cuhk.hk>
Sender: news@sparc15.cs.cuhk.hk
Organization: Engineering Faculty, The Chinese U. of Hong Kong
X-Newsreader: TIN [version 1.2 PL2]
Date: Tue, 9 Nov 1993 14:37:54 GMT
Lines: 37

Alan Miller
Bio-Medical Engineering Lab
Electronics Engineering
Chinese University of Hong Kong

We are small although quite active Bio-Medical Enghineering lab. As part 
of my PhD I have to model the acoutical shock wave propagation in human
tissue with a view to detection of disease. What has this to do with speach?

Well I want to take the output time series data that I get (the response
of the system to an acoustical impulse) and assume I have a transfer
function H
with a dirac delta input function. I then want to be able to find
the transfer function. And then look at the effect of not using a dirac
delta say sine input to the possitions of the pole/zero locus.
Yes its Linear Predictive Spectral Estimation or
the Maximum Entropy problem which I know is used in speach processing
(LPC10) This is only a very small aspect of my work and
it may not even work since my system is non-linear, time VARIANT,
inhomogenous and has no information available about the charactersitics
of the tissue groups. It may be that a linear model may suffice or it
may be that it will be totally useless. Either way I can not afford to
waste 5 months writing suitable routines for everything I need to do.

Therefore can anyone tell me where to find good quality DSP software for
sun SPARC that will have a good quality Linear Predictive Spectral
Estimation section. Or code that will take a complex waveform (like
speach) and map a fractal onto it (Fractal interpolation etc??). What
about other advanced analysis of time series data are the group familiar
with that I do not know of?


Any help, comments or general suggestions would be thnkfully received.
regards
Alan Miller.


