Speech analysis and synthesis apparatus

Abstract

ABSTRACT OF THE DISCLOSURE Disclosed is a speech analysis and synthesis apparatus having an analysis unit adapted to quantize the characterizing parameters of a speech signal at mutually different quantization steps depending on whether the speech signal represent a voiced sound or an unvoiced sound, and a synthesis unit adapted to decode the quantization output from the analysis unit to repro-duce the characterizing parameters. The unit subtracts a short-term mean value of environment noise from the short-term mean value of the sum of the ambient noise and the speech signal to provide the short-term mean value of the speech signal. The discrimination between the voiced and unvoiced sounds employed in the analysis unit nonlinearly converts decision parameters which have extremely different variance of occurrence rate dis-tribution characteristics between voiced and unvoiced sound used for the voiced-unvoiced decision equation, and analyzes the mixture of the environment-voice-representing signal and known voiced or unvoiced sound to determine the coefficient and thresh-old values of the voiced-unvoiced decision equation, thereby to determine the voiced or unvoiced sound based on the coefficients and threshold values. The apparatus requires reduced amount of transmission information without degrading the quality of synthesized speech sound.