Speech Database for Speech Analysis

Ganga Banavath, Sreedhar Potla
unpublished
Speech processing is becoming an emerging technology which enables interaction of human beings with machines. We are considering a standard length of speech sample which gives higher performance in speech processing. Our aim is to describe an efficient technique or method resulting effective speech processing applications. Performance of a classifier is a function depends on the length of speech sample, environment etc. This work is carried out using Mel frequency cepstral coefficients (MFCC)
more » ... ong with Gaussian Mixture Model (GMM) classifier. Results are evaluated on our own database comprising five South Indian Languages. The higher recognition performance is observed to be 82.6531% for 10seconds speech, 60.2041% for 5seconds speech and 57.1341% for 3seconds speech.
fatcat:y2l73m37pfg3birrekko2txowa