Optimizing gain codebook of LD-CELP

Author

Zhang, Gang ; Xie, Keming ; Zhang, Xueying ; Huangfu, Liying

Author_Institution

Taiyuan Univ. of Technol., China

Volume

2

fYear

2003

fDate

6-10 April 2003

Abstract

The paper combines the Jayant adaptive quantization (Jayant, N.S., Bell Syst. Tech. J., vol.52, p.1119-44, 1973) and the gain prediction of G.728 to propose a scheme suitable for designing the gain codebook of LD-CELP, in which the exact gain value is adaptively predicted and adaptively quantized. In this method, for adaptive quantization gain of 3 bit and 4 bit, the SNR has been increased by 0.5 dB and 6 dB, respectively, compared to that of G.728. If 4-bit adaptive quantization and the shape codebook of 64 codevectors are adopted, the complexity of G.728 could be reduced by 20% while SNR has 1 dB improvement compared with G.728. When the order of the synthesis filter is reduced from 50 to 30, the SNR is not changed and the complexity of G.728 is decreased by 40%.

Keywords

computational complexity; filtering theory; linear predictive coding; optimisation; quantisation (signal); speech coding; CELP; G.728 gain prediction; Jayant adaptive quantization; SNR; adaptive prediction; codevectors; complexity; gain codebook optimization; shape codebook; speech coding; synthesis filter; Adaptive filters; Decoding; Delay; Equations; Optimization methods; Performance gain; Quantization; Shape; Speech coding; Speech synthesis;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). 2003 IEEE International Conference on

ISSN

1520-6149

Print_ISBN

0-7803-7663-3

Type

conf

DOI

10.1109/ICASSP.2003.1202316

Filename

1202316