Speech processing involves segmenting input digital voice signal, transforming data blocks into frequency domain, weighting and feeding to low rate voice codec in time domain

摘要

The method involves segmenting an input digital voice signal into data blocks, successively transforming the blocks into the frequency domain, multiplying the resulting frequency components by time varying frequency dependent weighting factors depending on the last weighting factor used and feeding the thus weighted frequency components a low rate voice codec after transformation back into the time domain. AN Independent claim is also included for the following: a noise suppression arrangement.