Sign in
Search speechcoding
Subscribe to speechcoding
speechcoding by Keywords
ACELP | ADPCM | AMBE | AMR | AMR-NB | CELP | Codebook | DTMF | G.723 | G.726 | G.729 | GSM | Interpolation | LPC | LSF | LSP | MELP | PCM | Perceptual | Pitch | PSOL | QCELP | Quantization | SMV | VAD | Vocoder
Discussion Groups
Discussion Groups | Speech Coding | perceptual..
Technical discussions related to Speech Coding (all itu and other vocoders, ACELP, CELP, AMR, etc)
perceptual.. - Author Unknown - Apr 19 12:32:00 2004
|
hi all,
Can anyone tell me whatz the use of "PERCEPTUAL WEIGHTING FILTER" ?
why its so necessary in speech coding?
thanx in advance.
Mithun G N
|
(You need to be a member of speechcoding -- send a blank email to speechcoding-subscribe@yahoogroups.com )
Re: perceptual.. - B James - Apr 20 8:19:00 2004
|
Hi,
Analysis by synthesis is a coding scheme wherein error mimimization is performed
on difference signal ( difference between original speech signal & synthesized speech
samples ).
Error criterion like "mean square error" is just not sufficient in case of speech coding
& what is required is error criterion having more sympthy with human hearing (or perception
system), so we add a block of perceptual weighting filter.
perceptual weightng filter has frequency reponse which in effect flattens
the spectrum (inverted version of formant filter).
For details you can refer "Digital speech" by Kondoz
regards,
Mahesh
m...@yahoo.com wrote:
Yahoo! India Matrimony: Find your partner online. |
|
|
(You need to be a member of speechcoding -- send a blank email to speechcoding-subscribe@yahoogroups.com )
Re: perceptual.. - Shree Jaisimha - Apr 21 6:10:00 2004
|
Mithun
Perceptual coding is based on the frequency response of the ear.
The best way to learn about things instead of getting a book is the net.
And if you do get a book get the one by the acoustics guru J.F Moore ....please kindly
check this up as its been a long time since I touched this BIBLE. If I mistook the author name
please do a quick search on this also.
In the meantime, try doing a quick search on terms like temporal masking, psychoacoustics
and audio compression. Many audio compression algos are based on this assuming you have a
bandwidth ranging from 20 to 20 kHz.
For speech which in narrowband is bandlimited to 3.4 and wideband to 7 kHz you can use the
similar concepts.
basically in a nutshell perceptual coding means that you can afford to quantize as long as
you are below the noise level. In speech you have resonant modes termed as formants. Here you
can afford typically more compression and still get away with it with regards to the human ear
not perceiving it.
My friend Arijit is pretty much into this field and you can ask him more details, am
ccing his email herein.
S.Jaisimha
hi all,
Shree Jaisimha
|
(You need to be a member of speechcoding -- send a blank email to speechcoding-subscribe@yahoogroups.com )