The Nonlinear Time Alignment Model for Speech Recognition System

Abstract

We present the new nonlinear time alignment model, which is much faster than widely accepted DTW algorithms. This work has been started with the aim of finding suitable time alignment algorithm and features, for Marathi (Language spoken in the state of Maharashtra, India) word Speech Recognition System. Proposed algorithm shows comparable or better recognition efficiency than widely accepted algorithms and is robust to end point variations. In this work, vocabularies are: (1) 46 isolated monosyllabic confusing Marathi alphabets and (2) 46 non-confusing names of the persons. The features used are LPC, LFCC and MFCC. For the confusing word vocabulary, the proposed algorithm proved to be best showing maximum recognition efficiency of 89.13% and second best is Itakura's DTW algorithm with maximum recognition efficiency of 86.96%. LFCC with Itakura's DTW algorithm shows poor performance with maximum recognition efficiency of 13.40%. But LFCC with proposed algorithm shows comparable results for non-confusing vocabulary.

Indexing terms:

• DTW
• Confusing words
• Speech recognition system

Additional information

Notes on contributors

D D Doye

D D Doye received his BE (Electronics) degree in 1988 and ME (Electronics) degree in 1993, from SGGS College of Engineering and Technology, Vishnupuri, Nanded (MS). Presently, he is working as Assistant Professor in Department of Electronics and Computer Science and Engineering, SGGS College of Engineering and Technology, Vishnupuri, Nanded. His field of research is Speech Recognition.

T R Sontakke

T R Sontakke received his BE (Electrical Engineering) from Government College of Engineering, Aurangabad M Tech from VRCE Nagpur and PhD from IIT Bombay, Mumbai. He worked at TTTI Bhopal for a period of ten years. He is currently working as Professor in Electronics with additional charge of Principal, SGGS College of Engineering and Technology, Vishnupuri, Nanded. He has been the recipient of two Gold Medals from Institution of Engineers, India for his research papers. His research fields are Speech Recognition, Neural Networks and Image Processing.

Smita Nagtode

Smita Nagtode received his BE (Electronics and Telecommunication Engineering) from Bapuraoji Deshmukh College of Engineering Wardha (MS) in May/June 1998. She has received her ME (Electronics with specialization in Computer) degree from SGGS College of Engineering and Technology, Vishnupuri, Nanded in May 2000. Presently, she is working as Lecturer in department of Electronics, GH Raisoni College of Engineering and Technology, Digdoh Hills, Hingana Road, Nagpur.