Hi , I need some help regarding getting Cepstral coeffs from LPC coeffs. I have found 2 variation for the algo to do this. 1) One algo (given in fundamentals of Speech recogntion by Rabiner/Juang) says that the initial cepstral coeff is the residual gain term as calculated in the LPC phase. 2) Another algo that I found says that the first cepstral coeff is the first LPC coeff. I have a confusion regarding this. Will somebody please explain this? Thanks and reagards, Abhijeet Joshi Yahoo! India Insurance Special: Be informed on the best policies, services, tools and more. |
|
Help regarding Cepstral Coeffs (LPCC)
Started by ●February 23, 2004
Reply by ●February 24, 20042004-02-24
Hi , I found this information about cepstral coeffs on a site. "The zeroth cepstral coefficient cannot be computed directly with LPC analysis, but since it represents the overall energy in the spectrum, the LPC cepstrum coefficients are usually augmented with log-energy to replace ." I have a some questions about this: 1) Is this the error value obtained during LPC coeffs calculation? 2) Is this the zeroth autocorrelation coeff which represents the energy of the frame? Thanks and reagards, Abhijeet Joshi |