My understanding of it is this:
The hetro utility breaks the sound down into its component sinusoids, with
user-specified parameters controlling base frequency, number of partials,
number of time breakpoints. It writes an analysis file that is then used by
adsyn to control a bank of sine oscillators. The important thing here is
that adsyn is an oscillator bank, not filters.

In my experience adsyn/hetro works best with slowly varying sounds.
Transients, fricatives and plosives are difficult to reproduce, so accurate
speech synthesis is not too convincing. For example, I once analyzed and
resynthesized a bee in flight, that was very difficult to tell from the
original. I also tried it with a nightingale song. Less convincing but still
in the range of "reasonable facsimile."

hYdra (http://members.aol.com/additiv/) is a nice Windows utility for
editing the hetro analysis file prior to resynthesis.


Hope this helps.

-David.

> -----Original Message-----
> From: Sean Costello [mailto:costello@seanet.com]
> Sent: Monday, February 22, 1999 11:42 AM
> To: csound@maths.ex.ac.uk
> Subject: Hetro (was Re: Hetero)
> 
> 
> Hi all:
> 
> I meant to say "hetrodyne" and "hetro" instead of hetero.  My brain
> hurts today.
> 
> Sean
> 
> Sean Costello wrote:
> > 
> > Hi all:
> > 
> > Anyone have any information on heterodyne analysis besides what is
> > available in the Csound manual?  It seems very close to the 
> theoretical
> > description of the phase vocoder (when described as a 
> series of bandpass
> > filters, as opposed to the FFT-based version).  Does it have any
> > advantages to the phase vocoder, or is it just an older routine? (It
> > seems very close to an analysis program James Beauchamp 
> describes in a
> > computer music book I have from 1969).
> > 
> > Thanks,
> > 
> > Sean Costello