Pitch Detection of Speech Synthesis by Using Matlab

Abstract

In speech synthesis, machine is developed which can accept text and convert into natural sounding speech. Applications of speech synthesis include speech output from computers, reading machine for the visually challenged people. The difference between text to speech synthesizer and any other talking machine (e.g., cassette player) is, it could be trained for any speaker's voice in a fully automatic way. Three main approaches to speech synthesis: articulator synthesis, formant synthesis, and concatenate synthesis. I am carrying out with concatenate synthesis approach. In addition, text-to-speech (TTS) conversion system based on time-domain pitch-synchronous overlap-add (TD-PSOLA) method, has been employed to perform prosody (includes pitch, duration of a speech) modification. 7 To assure good quality of synthetic speech accurate estimation of pitch-period and pitch-marks are necessary for pitch modification. Pitch marking is divided into two tasks; pitch detection and location determination. LPF and some nonlinearity are being used for pitch- detection; peak-valley decision method is used to determine the appropriate parts of speech for used in pitch- mark estimation. In each pitch period, two possible peaks/valleys are searched and one dynamic programming is run to obtain pitch-mark.