Wavelet Codebook Design using Genetic Algorithm for Text Independent Speaker Recognition

S.-Y. Lung (Taiwan)

Keywords

Speaker recognition, Wavelet transform, Genetic Algorithm, Codebook design

Abstract

This paper presents text-independent speaker identification results for telephone speech. A speaker identification system based on wavelet transform (WT) derived from codebook design using genetic algorithm (GA) is proposed. We have combined genetic algorithm (GA) and the wavelet vector (WV) algorithm to avoid typical local minima for speaker data compression. Identification accuracies of 95% were achieved for 100 Mandarin speakers.

Important Links:

• DOI:
• From Proceeding (520) Signal Processing, Pattern Recognition, and Applications - 2006

Go Back