Re: Voice processing technology

[donley@blake.acs.washington.edu (E. Donley Olson)]

In article <3801@ur-cc.UUCP> boris@prodigal.psych.rochester.edu writes:
>
>Ever since I heard Laurie Anderson I've been wondering... how exactly
>do they do this filtering to change the sound of someone's voice, or
>to make it sound like 3 people singing at once...?
>

John Drukman has answered some of this...  regarding vocoders and such.
Harmonizers are strange little boxes.  I beleive that the harmonizers may
work by digitally sampling (or otherwise) the incomming sound, ie Laurie's
voice, and then playing it back "sped up" almost instantaneously... The
problem with this method is that you have to "drop" bits of the sound
when you do this because when you play it back "sped up" it takes less time
for the sound to occur and when you play it back "slowed down" it takes
too MUCH time to play it back.  The solution found in tape decks that use
this technique is to sample very quickly and to "cut out" the pieces that
are extra, or stick in an extra sample every period if the voice is sped up.
This is the least likely method because the  "cuts" leave annoying glitches
in the sound.

The other possible way this is accomplished is by doing some sort of frequency
counting on the incomming sound and then dividing the frequency by a certain
amount (or multiplying to make her sound like Dolly Parton).  I once built
a divider of this sort, but it was rather crude...  But it might be what
they do...  People have been able to digitize the human voice into square
waves, so why not...
I would NOT expect that harmonizers do complete spectral analysis
on samples in real time -- even the Fairlight doesn't do that!

Any other possibilities?
  - Eo