Vector quantization and signal compression
Vector quantization and signal compression
A novel dimension conversion for the quantization of SEW in wideband WI speech coding
ICIC'05 Proceedings of the 2005 international conference on Advances in Intelligent Computing - Volume Part I
Hi-index | 0.00 |
A speech coding algorithm for application to the Japan digital cellular half-rate coder has been developed based on the Multi-Band Excitation (MBE) algorithm using 3.0Kbps. and the target performance is at the same level as a full-rate coder operating at 6.7Kbps. This paper mainly describes the quantization scheme for the spectral envelope of the MBE coder parameters. New features of the scheme include an efficient dimension converter, inter-frame coding of spectral envelope vectors, and Multi-Stage Vector Quantization with Interpolative Vector Quantization. A simplified model of voiced/unvoiced division and a method of detecting/ manipulating background noise have also been devised for MBE analysis. Informal listening tests indicate that. while the quality is slightly lower than that of a full-rate coder, it is, nevertheless, sufficiently high for digital telephony.