VOICED SOUND/UNVOICED SOUND DECISION METHOD AND APPARATUS THEREFOR AND SPEECH ENCODING METHOD

摘要

PROBLEM TO BE SOLVED: To make voiced sound/unvoiced sound (V/UV) decision with a simple algorithm and with high accuracy by collectively judging the respective input parameters for V/UV decision. SOLUTION: As the parameters for the voiced sound/unvoiced sound decision relating to an input speech signal, the frame average energy lev of the input speech signal, normalization self-correlation peak value rOr, spectral similarity pos, zero corss number nZero and pitch lag pch are supplied to input terminals 11 to 15. When these parameters are defined as (x), the parameters are respectively converted by the sigmoid functions g(x) respectively expressed by the equation g(x)=1/(1+exp(-(x-b)/a)) (where a, b are constants) by function calculation circuits 31 to 35. The voiced sound/unvoiced sound decision is executed by a V/UV decision circuit 26 by using these converted parameters.