Dear all, I would like to know if for the floating point versions of the NB-AMR and WB-AMR speech coders, there exist a standardized way of testing the implementation. I mean a set of input and reference output vectors (as in the fixed-point versions), and a meaningful way to compare the output vectors of our implementation, versus the output reference vectors (e.g. average SNR threshold, etc.). I am also interested in similar testing procedures for other standard coders in floating point version. I have seen somenthing like this in the G.723.1 floating point, but the information was not very enlightening for me :-( Thanks in advance for any hint you could give me, Regards, Sara |