2012Unpublished venueRequires access

Comparative study on vector quantization codebook generation algorithms for wideband speech coding

Kavya Manohar, B. Premanand

Open publisher page 4 citations

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.

About this research paper

What this paper is about

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.

Why it matters

OpenAlex reports 4 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available 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.

Key concepts: Codebook, Linde–Buzo–Gray algorithm, Code-excited linear prediction, Vector quantization, Vector sum excited linear prediction, Speech coding, Linear predictive coding, Computer science

Related papers

Back to paper searchBrowse research topicsOriginal source
Comparative study on vector quantization codebook generation algorithms for wideband speech coding — Research Paper | ScholarLens