Digital speech processing system having reduced encoding bit requirements

摘要: A digitized speech signal is divided into sections and each section analyzed by the linear prediction method to determine coefficients of a sound formation model, volume parameter, information concerning voiced or unvoiced excitation period vocal band base frequency. In order improve quality without increasing data rate, redundance reducing coding parameters effected. The performed in blocks two three adjacent sections. first are coded complete form, those other differential form part not at all. average number bits required per reduced compensate for increased so that overall rate increased.