Speech Recognition Challenges by using Neural Network Approaches

Abstract

Speech technology and frameworks in user computer dealings have seen a steady and notable progress over the past twenty years. Presently, speech technologies are accessible on the market for an infinite though appealing set of functions. Such technologies allow devices to answer accurately and dependably to people’s voices, and give helpful and important services. Current studies focus on exploiting frameworks that will be significantly stronger against changes in surroundings, user and language. Thus current studies mostly concentrate on ASR frameworks having a consequent glossary that enable speaker autonomous process comprising fluid speech in dissimilar tongues. This article provides a summary of the speech identification framework and its current development. The basic aim of this article is to contrast and recapitulate some of the popular techniques employed in different levels of speech identification framework. Key Words: Speech Recognition; Feature Extraction; MFCC; LPC; Hidden Markov Model; Neural Network; Dynamic Time Warping.