SPEECH ENCODER, SPEECH DECODER, SPEECH ENCODING METHOD, SPEECH DECODING METHOD, SPEECH ENCODING PROGRAM AND SPEECH DECODING PROGRAM Russian patent published in 2013 - IPC G10L21/02 

FIELD: information technology.

SUBSTANCE: signal linear prediction coefficient presented in the frequency domain is obtained by performing frequency analysis with linear prediction using a covariation technique or an autocorrelation technique. Once the filter power of the obtained linear prediction coefficient is corrected, the signal is frequency filtered using the corrected coefficient, thereby forming a signal time envelope.

EFFECT: reduced pre-echo and delayed echo, which improves the subjective quality of the decoded signal without considerable increase in bit rate in a bandwidth expansion method in the frequency domain, represented by a spectral band replication method.

2 cl, 50 dwg