y[n] is needed to calculate the minimum error. y[n] is filtered by h[n]-synthesis filter to get synthesised speech. The difference of y[n] and original speech i.e error is what algorithm is trying to minimize. Hope this help. Nayan Gaywala --- In , "Sameer Kibey" <sameer@t...> wrote: > hi all > > in the AMR standard, a signal called y[n] is used. it is the "filtered adaptive codebook vector" and is defined as: > y[n] = v[n]*h[n] (read convolution of v[n] and h[n]), where > v[n] => adaptive codebook vector and > h[n] => impulse response of teh weighted synthesis filter. > > i am unable to understand what y[n] actually represents. what is the need of filtering the content of the adaptive codebook with the short term predictor? can anyone please explain the significance of this signal y[n] ? > > thanks > Sameer. |