FIELD: information technology.
SUBSTANCE: transmission of information on parametres of the forming model of a vocoder with linear prediction is excluded from the frame when coding successive vocalised segments. Information on given parametres is transmitted only for the first vocalised segment in the sequence and stored unchanged for the rest. The number of such segment must not exceed four. Parametres of the forming model are calculated again for the next fifth vocalised segment. When coding noise-like and transition segments, parametres of the synthesising system are calculated on each speech signal quasistationarity segment. In steady-state conditions of the sequence of vocalised segments of speech signal quasistationarity during coding, an information signal confirming that a segment belongs to a vocalised type is transmitted over a transmission channel, as well as an actuating signal calculated based on the procedure for analysis through synthesis, and its amplification coefficient which controls power of the speech signal at the output of the synthesis filter.
EFFECT: lower channel transmission capacity required for vocoders with linear prediction in the class of systems with variable transmission speed.
4 dwg
