[speex-dev] project 'Sphinx' kicked off

Jean-Marc Valin jean-marc.valin at hermes.usherb.ca
Thu Jun 19 20:55:51 PDT 2003

> Have you tried working with fixed point IIR filters?  Stable or not, the
> frequency response suffers.   Maybe it will still be good enough.

I think it'll be good enough. Keep in mind that speech codecs have musch
looser quality requirements for these things than, say, Vorbis.

> Are the coefficients for the LPC or other IIR filters dependent on the data?
> I thought that they were derived from the modes and submodes defined in
> the various flavors.
> If the coefficients are data-dependent, that makes things considerably
> more difficult.

Of course it's dependent on the data. For every frame, the encoder
estimates LPC coefficients and transmit them (as LSP's) to the decoder.
The decoder uses these coefficients (that change for each sub-frame
because of filter interpolation) to reconstruct the signal. Regardless,
it's really no big deal.


Jean-Marc Valin, M.Sc.A.
LABORIUS (http://www.gel.usherb.ca/laborius)
Université de Sherbrooke, Québec, Canada

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 190 bytes
Desc: Ceci est une partie de message numériquement signée
Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20030619/415d50ac/signature-0001.pgp

More information about the Speex-dev mailing list