Word and Speaker Recognition using Concept of MFCC and Correlation Coefficients |
Author(s): |
Akshay Ramesh , Mukesh Patel School of Technology Management and Engineering, NMIMS (Deemed-to-be University); Jaykumar Chaudhary, Mukesh Patel School of Technology Management and Engineering, NMIMS (Deemed-to-be University); Sayantan Chakraborty, Mukesh Patel School of Technology Management and Engineering, NMIMS (Deemed-to-be University); Nieves Crasto, Mukesh Patel School of Technology Management and Engineering, NMIMS (Deemed-to-be University) |
Keywords: |
Word and Speaker Recognition, MFCC, Correlation Coefficients |
Abstract |
Speech and speaker recognition has permeated into everyday technology. Most recognition systems are based on machine learning techniques employing deep neural networks for the classification. This paper deals with providing a simple solution to yes/no speech classification using Fourier Transform and histogram based thresholding. Most automatic speech recognition system use Mel Frequency Coefficients (MFCCs) are features for classification. In this paper instead of using MFCCs as features directly, correlation coefficients between MFCCs of different frames are used as features and classification is carried out using Euclidean distance based template matching. An accuracy of 85% and 98.375% was achieved yes/no classification and speaker recognition respectively. |
Other Details |
Paper ID: NCTAAP048 Published in: Conference 4 : NCTAA 2016 Publication Date: 29/01/2016 Page(s): 207-209 |
Article Preview |
Download Article |
|