A Speaker Identification System Using MFCC Features with VQ Technique

A. Zulfiqar, Muhammad Aslam, A. M. Martinez Enriquez

Research output: Chapter in Book/Report/Conference proceedingConference contribution

36 Citations (Scopus)

Abstract

The performance of speaker identification systems has improved due to recent advances in speech processing techniques but there is still need of improvement in term of text-independent speaker identification and suitable modelling techniques for voice feature vectors. It becomes difficult for person to recognize a voice when an uncontrollable noise adds in to it. In this paper, feature vectors from speech are extracted by using mel-frequency cepstral coefficients and vector quantization technique is implemented through Linde-Buzo-Gray algorithm. Two purposeful speech databases with added noise, recorded at sampling frequencies 8000 Hz and 11025 Hz, are used to check the accuracy of the developed speaker identification system in non-ideal conditions. An analysis is also provided by performing different experiments on the databases that number of vectors in VQ codebook and sampling frequency influence the identification accuracy significantly.
LanguageEnglish
Title of host publication2009 Third International Symposium on Intelligent Information Technology Application
Pages115-118
Number of pages4
Publication statusPublished - 31 Dec 2009

Fingerprint

Identification (control systems)
Sampling
Speech processing
Vector quantization
Experiments

Keywords

  • cepstral analysis
  • signal sampling
  • speaker recognition
  • vector quantisation
  • speaker identification
  • MFCC feature
  • speech processing
  • voice feature vector
  • mel-frequency cepstral coefficient
  • vector quantization
  • Linde-Buzo-Gray algorithm
  • speech database
  • VQ codebook
  • sampling frequency
  • frequency 8000 Hz
  • frequency 11025 Hz
  • Mel frequency cepstral coefficient
  • Cepstral analysis
  • Vector quantization
  • Noise figure
  • Sampling methods
  • Hidden Markov models
  • Working environment noise
  • Speech recognition
  • Speech enhancement
  • Spatial databases
  • Speaker identification
  • MFCC
  • VQ

Cite this

Zulfiqar, A., Aslam, M., & Martinez Enriquez, A. M. (2009). A Speaker Identification System Using MFCC Features with VQ Technique. In 2009 Third International Symposium on Intelligent Information Technology Application (pp. 115-118)
Zulfiqar, A. ; Aslam, Muhammad ; Martinez Enriquez, A. M. / A Speaker Identification System Using MFCC Features with VQ Technique. 2009 Third International Symposium on Intelligent Information Technology Application. 2009. pp. 115-118
@inproceedings{351871c6f5d94dd09626a352f0297cd0,
title = "A Speaker Identification System Using MFCC Features with VQ Technique",
abstract = "The performance of speaker identification systems has improved due to recent advances in speech processing techniques but there is still need of improvement in term of text-independent speaker identification and suitable modelling techniques for voice feature vectors. It becomes difficult for person to recognize a voice when an uncontrollable noise adds in to it. In this paper, feature vectors from speech are extracted by using mel-frequency cepstral coefficients and vector quantization technique is implemented through Linde-Buzo-Gray algorithm. Two purposeful speech databases with added noise, recorded at sampling frequencies 8000 Hz and 11025 Hz, are used to check the accuracy of the developed speaker identification system in non-ideal conditions. An analysis is also provided by performing different experiments on the databases that number of vectors in VQ codebook and sampling frequency influence the identification accuracy significantly.",
keywords = "cepstral analysis, signal sampling, speaker recognition, vector quantisation, speaker identification, MFCC feature, speech processing, voice feature vector, mel-frequency cepstral coefficient, vector quantization, Linde-Buzo-Gray algorithm, speech database, VQ codebook, sampling frequency, frequency 8000 Hz, frequency 11025 Hz, Mel frequency cepstral coefficient, Cepstral analysis, Vector quantization, Noise figure, Sampling methods, Hidden Markov models, Working environment noise, Speech recognition, Speech enhancement, Spatial databases, Speaker identification, MFCC, VQ",
author = "A. Zulfiqar and Muhammad Aslam and {Martinez Enriquez}, {A. M.}",
year = "2009",
month = "12",
day = "31",
language = "English",
isbn = "978-0-7695-3859-4",
pages = "115--118",
booktitle = "2009 Third International Symposium on Intelligent Information Technology Application",

}

Zulfiqar, A, Aslam, M & Martinez Enriquez, AM 2009, A Speaker Identification System Using MFCC Features with VQ Technique. in 2009 Third International Symposium on Intelligent Information Technology Application. pp. 115-118.

A Speaker Identification System Using MFCC Features with VQ Technique. / Zulfiqar, A.; Aslam, Muhammad; Martinez Enriquez, A. M.

2009 Third International Symposium on Intelligent Information Technology Application. 2009. p. 115-118.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A Speaker Identification System Using MFCC Features with VQ Technique

AU - Zulfiqar, A.

AU - Aslam, Muhammad

AU - Martinez Enriquez, A. M.

PY - 2009/12/31

Y1 - 2009/12/31

N2 - The performance of speaker identification systems has improved due to recent advances in speech processing techniques but there is still need of improvement in term of text-independent speaker identification and suitable modelling techniques for voice feature vectors. It becomes difficult for person to recognize a voice when an uncontrollable noise adds in to it. In this paper, feature vectors from speech are extracted by using mel-frequency cepstral coefficients and vector quantization technique is implemented through Linde-Buzo-Gray algorithm. Two purposeful speech databases with added noise, recorded at sampling frequencies 8000 Hz and 11025 Hz, are used to check the accuracy of the developed speaker identification system in non-ideal conditions. An analysis is also provided by performing different experiments on the databases that number of vectors in VQ codebook and sampling frequency influence the identification accuracy significantly.

AB - The performance of speaker identification systems has improved due to recent advances in speech processing techniques but there is still need of improvement in term of text-independent speaker identification and suitable modelling techniques for voice feature vectors. It becomes difficult for person to recognize a voice when an uncontrollable noise adds in to it. In this paper, feature vectors from speech are extracted by using mel-frequency cepstral coefficients and vector quantization technique is implemented through Linde-Buzo-Gray algorithm. Two purposeful speech databases with added noise, recorded at sampling frequencies 8000 Hz and 11025 Hz, are used to check the accuracy of the developed speaker identification system in non-ideal conditions. An analysis is also provided by performing different experiments on the databases that number of vectors in VQ codebook and sampling frequency influence the identification accuracy significantly.

KW - cepstral analysis

KW - signal sampling

KW - speaker recognition

KW - vector quantisation

KW - speaker identification

KW - MFCC feature

KW - speech processing

KW - voice feature vector

KW - mel-frequency cepstral coefficient

KW - vector quantization

KW - Linde-Buzo-Gray algorithm

KW - speech database

KW - VQ codebook

KW - sampling frequency

KW - frequency 8000 Hz

KW - frequency 11025 Hz

KW - Mel frequency cepstral coefficient

KW - Cepstral analysis

KW - Vector quantization

KW - Noise figure

KW - Sampling methods

KW - Hidden Markov models

KW - Working environment noise

KW - Speech recognition

KW - Speech enhancement

KW - Spatial databases

KW - Speaker identification

KW - MFCC

KW - VQ

UR - https://ieeexplore.ieee.org/document/5369090

M3 - Conference contribution

SN - 978-0-7695-3859-4

SP - 115

EP - 118

BT - 2009 Third International Symposium on Intelligent Information Technology Application

ER -

Zulfiqar A, Aslam M, Martinez Enriquez AM. A Speaker Identification System Using MFCC Features with VQ Technique. In 2009 Third International Symposium on Intelligent Information Technology Application. 2009. p. 115-118