Text independent automatic speaker recognition using selforganizing maps

Author

Mafra, Alexandre Teixeira ; Simões, Marcelo Godoy

Author_Institution

Escola Politecnica, Sao Paulo Univ., Sao Paulo, Brazil

Volume

3

fYear

2004

Firstpage

1503

Abstract

This work presents one implementation of an automatic speaker recognition system, based on selforganizing map (SOM) neural networks. The voice of each speaker is modeled by a SOM, trained to specialize in the quantization of feature vectors (MFCCs) extracted from his voice. When a test sample is presented, it is quantized by all SpMs, that compete for the speaker: the SOM with smallest quantization error defines the speaker. The system was tested on a speaker identification task over a 14 speaker set, with phrases from three phonetically balanced sets and one variable answer set. The results comprovate the method´s efficiency.

Keywords

cepstral analysis; self-organising feature maps; speaker recognition; vector quantisation; MFCC; SOM neural network; VQ; automatic speaker recognition; mel-frequency cepstral coefficient; selforganizing map neural network; speaker identification task; vector quantization error; Artificial neural networks; Data mining; Feature extraction; Frequency; Hidden Markov models; Neural networks; Self organizing feature maps; Speaker recognition; Testing; Vector quantization;

fLanguage

English

Publisher

ieee

Conference_Titel

Industry Applications Conference, 2004. 39th IAS Annual Meeting. Conference Record of the 2004 IEEE

ISSN

0197-2618

Print_ISBN

0-7803-8486-5

Type

conf

DOI

10.1109/IAS.2004.1348670

Filename

1348670