[music-dsp] Killer speech noise reduction

[James Chandler Jr.]

It is said that govt agencies persue secrecy for secrecy's sake, even if
there is no particular need for it on some cases.

I can understand how advanced technology on stuff like image recognition or
data encryption would be strategically advantageous to keep secret. But it
is more difficult to imagine how technology to pull voice out of a noisy
signal would be so sensitive.

When customers inquire looking for programs to transcribe polyphonic audio
recordings, I usually say that with the current state of the art, it would
take an army of engineers and a defense contractor budget to write a really
good one (ferinstance, something strong enough to accurately transcribe the
score of a Beethoven symphony or whatever).

Dunno if that is really true, or how long it will remain true.

But perhaps real good speech extraction is another one of those tasks that
would require an army of engineers and a defense contractor budget (or
perhaps a Bill Gates budget (GRIN)).

JCJR

"In Shiva we Trust" (GRIN)

> If you want to dance to the music, you have to pay it to the piper.  If
you
> want no one else to dance to the music, you have to pay more to the piper
> and watch the piper.
>





dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links
http://shoko.calarts.edu/musicdsp/