Re: Neural Nets in MP3 Encoding

Mathew Hendry (scampi@dial.pipex.com)
Tue, 25 May 1999 04:24:32 +0200 (MET DST)

On 22 May 1999 13:38:16 GMT, A Ahmad <A2AHMAD@engmail.uwaterloo.ca>
wrote:

>I'm researching to develop a new encoder for the MPEG Audio Layer III
>format. The encoders need among other things a psychoacoustic models to
>get rid of the information about sounds that are masked by louder tones.
>I'm trying to determine whether a technique based on fuzzy logic and/or
>neural net principles would help. Any thoughts?

It depends on how far you want to take it, I suppose. If you already
have a reasonably good masking model, you could initialise the net to
represent that model, and then use listening tests to train and
improve it.

Your first step, then, would be to find a way of representing an
arbitrary, adjustable masking model as a neural network. I suspect
that time-domain masking (i.e. a given signal masks an earlier or
later signal) would be the trickiest to implement flexibly.

-- Mat.

############################################################################
This message was posted through the fuzzy mailing list.
(1) To subscribe to this mailing list, send a message body of
"SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at
(2) To unsubscribe from this mailing list, send a message body of
"UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com"
to listproc@dbai.tuwien.ac.at
(3) To reach the human who maintains the list, send mail to
fuzzy-owner@dbai.tuwien.ac.at
(4) WWW access and other information on Fuzzy Sets and Logic see
http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info
(5) WWW archive: http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html