Sixth European Conference on Speech Communication and Technology
(EUROSPEECH'99)

Budapest, Hungary
September 5-9, 1999

Source-Dependent Variable Rate Speech Coding below 3 KBPS

M. Stefanovic, A. Kondoz

Centre for Communication Systems Research (CCSR), University of Surrey, Guildford, Surrey, UK

This work addresses the need for high quality, very low bit rate speech compression algorithms that can be utilised in many forthcoming multimedia applications. The speech coding algorithm proposed in this paper is a variable rate system based on the adaptive source-driven frame length scheme. Maximum speech compression is achieved for long-term steady-state speech and non-speech (silence and unvoiced) conditions. In addition, shorter frame sizes are used to code those difficult-to-model speech transitions, thus improving the overall perceptual quality when compared with traditional fixed rate schemes. This codec may be used for implementing various voice communication systems, such as Voice Store and Forward systems and digital answering machines, or to augment low bit rate integrated digital packet networks.


Full Paper (PDF)   Gnu-Zipped Postscript

Bibliographic reference.  Stefanovic, M. / Kondoz, A. (1999): "Source-dependent variable rate speech coding below 3 KBPS", In EUROSPEECH'99, 1487-1490.