DocumentCode :
3535819
Title :
Comparative study on vector quantization codebook generation algorithms for wideband speech coding
Author :
Manohar, K. ; Premanand, B.
Author_Institution :
Dept. of ECE, Gov. Eng. Coll., Thrissur, India
fYear :
2012
fDate :
18-20 Dec. 2012
Abstract :
This paper presents a review and performance comparison on several vector quantization (VQ) codebook generation algorithms. The codebook generation algorithms discussed include K-means algorithm, Multi stage vector quantization (MSVQ) algorithm and Predictive split vector quantization (PSVQ)algorithm. The vectors under consideration are the Line Spectral Frequency (LSF) vectors obtained by transforming Linear Prediction Coefficients (LPCs) extracted during Code Excited Linear Prediction(CELP) coding of wideband speech. The study compares the algorithms based on their computational cost and the perceptual quality of resultant speech.
Keywords :
quantisation (signal); speech coding; CELP; K-means algorithm; MSVQ algorithm; PSVQ algorithm; VQ codebook generation algorithms; code excited linear prediction; computational cost; line spectral frequency; linear prediction coefficients; multistage vector quantization; perceptual quality; predictive split vector quantization; resultant speech; vector quantization codebook generation algorithms; wideband speech coding; Speech; Speech coding; Training; Vector quantization; Vectors; Wideband;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Green Technologies (ICGT), 2012 International Conference on
Conference_Location :
Trivandrum
Print_ISBN :
978-1-4673-2635-3
Type :
conf
DOI :
10.1109/ICGT.2012.6477952
Filename :
6477952
Link To Document :
بازگشت