Optimal quantization of LSP parameters

Author

Soong, Frank K. ; Juang, Biing-hwang

Author_Institution

AT&T Bell Lab., Murray Hill, NJ, USA

Volume

1

Issue

1

fYear

1993

fDate

1/1/1993 12:00:00 AM

Firstpage

15

Lastpage

24

Abstract

Two nonuniform aspects of the line spectrum pair (LSP) linear predictive coding (LPC) parameters are investigated, including nonuniform statistical distributions and spectral sensitivities of adjacent LSP frequency differences. Based upon these two nonuniform properties, a globally optimal scalar quantizer is designed for each differential LSP frequency. The design algorithm is dynamic programming based and minimization of a nontrivial data dependent spectral distortion is adopted as the optimality criterion. At 32 bits/frame, the new LSP quantizer achieves a 1-dB average log spectral distortion, a commonly accepted level for reproducing perceptually transparent spectral information. The quantization performance has also been shown to be robust across different speakers and databases

Keywords

linear predictive coding; optimisation; speech coding; vector quantisation; 32 bit; LPC parameters; LSP frequency differences; LSP parameters; LSP quantizer; average log spectral distortion; data dependent spectral distortion; databases; design algorithm; dynamic programming; globally optimal scalar quantizer; line spectrum pair; linear predictive coding; minimization; nonuniform properties; nonuniform statistical distributions; optimal quantisation; quantization performance; speakers; spectral sensitivities; Algorithm design and analysis; Databases; Dynamic programming; Frequency; Heuristic algorithms; Linear predictive coding; Minimization methods; Quantization; Robustness; Statistical distributions;

fLanguage

English

Journal_Title

Speech and Audio Processing, IEEE Transactions on

Publisher

ieee

ISSN

1063-6676

Type

jour

DOI

10.1109/89.221364

Filename

221364