Forums

G729 quantization of the gains

Started by shabnamso December 9, 2003
Hello
I would be thankful if anybody who has tested the G729A source code
by ITU can tell me if the QUA_GAIN function has been implemented
exactly as the
algorithm.
I try to run it on Code Composer Studio C6000 simulator with various
ITU test
vectors and everytime I get the error of division by a negative
number:'denom'
I appreciate your suggestions in this regard.
Shabnam




Hi,
The code is Implemented in the same way as the Spec,
But the problem is with the saturation.
Don't remove saturation for L_add, ie don't replace
L_add(a, b) with a+b. In C6000 you have sadd which
does saturation on addtion. This is the problem which
I think and not with the Spec to code mapping.

--- shabnamso <> wrote:
> Hello
> I would be thankful if anybody who has tested the
> G729A source code
> by ITU can tell me if the QUA_GAIN function has been
> implemented
> exactly as the
> algorithm.
> I try to run it on Code Composer Studio C6000
> simulator with various
> ITU test
> vectors and everytime I get the error of division
> by a negative
> number:'denom'
> I appreciate your suggestions in this regard.
> Shabnam >
>


=====

With Rgds
Chetan k.

__________________________________