Hi everyone, I am working on ITU Standard G.723.1, ie. speech coder(Hybrid coder)i have started the project one month back. I have almost designed the h/w for it. And in s/w part i have gone through th b/d of G.723.1, I am clear with most of the thing but i am unable to understand the functionality of MP-MLQ. Please help me in getting the MP-MLQ concept. Waiting for reply, Rajesh Yadav ME ETRX COMPUTER GCOE, Pune-5 |
|
MP-MLQ
Started by ●September 16, 2003
Reply by ●September 17, 20032003-09-17
hello Rajesh MP-MPLQ module in G.723.1 approximates the secondary excitation with the help of a few pulses. The job of this module is to find the locations, signs and the gain factors for these pulses, which in turn become the parameters to be encoded. please go through the topic on multi-pulse excitation (MPE) in the book by A. Kondoz. That would certainly help you understand the working of MP-MLQ in G.723.1. best regards, Sameer -----Original Message----- From: prajesh_etc [mailto:] Sent: Wednesday, September 17, 2003 3:50 AM To: Subject: [speechcoding] MP-MLQ Hi everyone, I am working on ITU Standard G.723.1, ie. speech coder(Hybrid coder)i have started the project one month back. I have almost designed the h/w for it. And in s/w part i have gone through th b/d of G.723.1, I am clear with most of the thing but i am unable to understand the functionality of MP-MLQ. Please help me in getting the MP-MLQ concept. Waiting for reply, Rajesh Yadav ME ETRX COMPUTER GCOE, Pune-5 |
|
Reply by ●October 13, 20032003-10-13
Hi everyone, I am working on ITU Standard G.723.1 Annex B, ie. speech coder have started the project. This coder is a type of Analysis-by-Synthesis, in this we use to minimise the error by Formant perceptual weighting filter. I have gone through the standard as well as source code. I am clear with most of the thing but i am unable to understand how this FPW minimise the error signal. Please help me in getting the FPW concept. Thanks and Regards Nihal |
|
Reply by ●October 14, 20032003-10-14
Hi, I think this was well explained before by Sameer Kibey (message number 580 and below), actually he talked about both harmincs and formants perceptual filters. You should read them. But I'd like to add some : If look at the spectrum of the weighting filter, it has the inverse of the speech spectrum shape (i.e the shape of the synthesis filter, but not exactly the same because of the use of expansion parameters gammas). The minimzation process tends to give a flat spectrum error coding. By using the perceptual filter, the weighted error is also flat (it's the role of the error minimzation), and since this error was shaped by the weighting filter, then the error evaluated in the non-weighted domain will have the inverse shape of the weighting filter (i.e shape of the speech but less prounouced because of the expansion parametres). Hope this will help. --- In , Nihal Ahmad <q_nihal@y...> wrote: > Hi everyone, > I am working on ITU Standard G.723.1 Annex B, ie. > speech coder have started the project. This coder is a > type of Analysis-by-Synthesis, in this we use to > minimise the error by Formant perceptual weighting > filter. I have gone through the standard as well as > source code. I am clear with most of the thing but i > am unable to understand how this FPW minimise the > error signal. > > Please help me in getting the FPW concept. > > Thanks and Regards > Nihal |