Redundant codes in ITU-T G729A?? | Speech Coding

Technical discussions related to Speech Coding (all itu and other vocoders, ACELP, CELP, AMR, etc)

Dear all, I'm a novice on speech codec and recently just started on implementation and optimization of the G729A codec. However, while doing a coverage checking of the encoder module, the following were noticed. 1. In Az_lsp module, when calculating the roots for sum & difference polynomial using linear interpolation, xint = xlow-ylow*(xhigh-xlow)/(yhigh-ylow) the program checks for condition when (yhigh-ylow) equals zero and equates xint to xlow. Is this portion of codes redundant since ylow is dependent on grid and grid varies for each computation of ylow? 2. In Gbk_presel module, condition for gcode0 <= 0 is checked. However, gcode0 is computed as 2 to the power of 14.(frac) in Gain_predict module and hence could never be less than or equals zero. Is the above condition (gcode0 <= 0) redundant as well? Any comments on the above would be very much appreciated. Thank you in advance! Regards, chin wee

(You need to be a member of speechcoding -- send a blank email to speechcoding-subscribe@yahoogroups.com )

Hi, The answer to ur first interpretation which is wrong is as follows, When we do the interpolation and if the value matches with the correct value, then we assign the correct value which happens to be xlow. Otherwise we do the calculation inorder to get a value as near as possible. This value is nothing but the position of the change in sign. With Rgds Chetan k.

(You need to be a member of speechcoding -- send a blank email to speechcoding-subscribe@yahoogroups.com )

Sign in

Search speechcoding

Subscribe to speechcoding

speechcoding by Keywords

Sponsor

Discussion Groups

Discussion Groups | Speech Coding | Redundant codes in ITU-T G729A??

Redundant codes in ITU-T G729A?? - chinwee_kwek - Aug 5 14:41:00 2002

Re: Redundant codes in ITU-T G729A?? - chetan k - Aug 6 10:58:00 2002