Speaker identification and Spoken word recognition in noisy background using artificial neural networks

Abstract

Generally Speech Recognition Systems are specific to speech/spoken word recognition or Speaker Identification/Verification. In this paper, An attempt has been made to find the better combination of Speech feature extraction and Artificial Neural Network Model for Speaker Identification combined with Spoken word recognition in general noisy back ground (i. e Home/Office environment). Different speech feature extraction techniques such as Mel Frequency cepstarl coefficients (MFCC), Perceptual Linear Prediction (PLP) Cepstral Coefficients and Gammatone Frequency Cepstral Coefficients (GFCC) in combination with two different Neural Network models such as Radial Basis Neural Networks and Learning Vector Quantization Neural Networks have been experimented. Three different test categories such as Spoken word recognition, Speaker Identification, and the combination of both speaker and spoken word recognition have been experimented for the above mentioned combinations. It is Suggested from the experiments that the combination of GFCC and Radial Basis Neural Networks gives the better recognition success rate in general noisy environment.