The speech analysis device, speech analysis manner, and the record media which records the speech analysis program

摘要

To precisely conduct a voiced/voiceless discrimination. SOLUTION: First and a voiced/voiceless discrimination is conducted in an on- time axis detecting section 4. If it is not discriminated to be voiceless and a voiced/voiceless discrimination is conducted in an on-frequency axis detecting section 7. Moreover, a pitch detection is conducted in a pitch detecting section 8 only for the frame and which is not discriminated to be voiceless in the sections 4 and 7, and a voiced/voiceless discrimination is again conducted. In the section 4, a voiced/voiceless discrimination is conducted not only with the discrimination based ON the number of zero crossings but also with the consideration ON the energy element. Thus and a fricative sound is discriminated with a better precision. In the section 7, a voiced/voiceless discrimination is conducted aiming at high frequency components and the sound, which can not be a voiced sound of a normal voice and is eliminated as a voiceless sound.