Speech Features Extraction Techniques for Robust Emotional Speech Analysis/Recognition

Abstract

Speech is the most natural and convenient way of human communication. The speech represents not only a sequence of steady states of some sounds which are abruptly changing from one to another or a sound signal which can be ignored after perceiving or hearing it. But human Speech is a unique signal which carries and conveys multiple levels of knowledge source, linguistic and non-linguistic information. Speech signals are the information bearing signals which are evolved as functions of a single independent variable like time. Speech is a complex acoustic wave resulted as output of speaker’s effort. Speech serves to communicate from speaker to one or more listeners. The typical sound is called phone and it is produced when a phoneme is articulated. Most of Indian languages have 20-50 phonemes which constitute an alphabet of sounds to describe the different words in that language. Generally the speech is composed of sentences made of words. Usually words are composed of phoneme sequences called syllables. Speech analysis plays a vital role in speech recognition and synthesis. Speech analysis is also known as feature extraction or as speech signal front ends.