Poonam Sharma and Abha Kiran Rajpoot, Sharda University, India

The objective of this work is to automatically segment the speech signal into silence, voiced and unvoiced regions which are very beneficial in increasing the accuracy and performance of recognition systems. Proposed algorithm is based on three important characteristics of speech signal namely Zero Crossing Rate, Short Time Energy and Fundamental Frequency. The performance of the proposed algorithm is evaluated using the data collected from four different speakers and an overall accuracy of 96.61 % is achieved.

Zero Crossing Rate, Short Time Energy, Fundamental Frequency, Cepstrum.