EUROSPEECH '93
Third European Conference on Speech Communication and Technology

Berlin, Germany
September 22-25, 1993

      

Fast Vector Quantization Using Neural Maps for CELP at 2400bps

Eduardo Lopez-Gonzalo, Luis A. Hernandez-Gomez

Dpto. SSR, ETSI Telecomunicacion, UPM, Ciudad Universitaria s/n, Madrid, Spain

The aim of the paper is to discuss fast vector quantization procedures for LPC parameters in CELP type coders at low bit rates. For this purpose, LPC representations will be based on the quantization properties of Self-Organizing Maps (SOM). These Maps provide an efficient representation of the LPC parameters that is exploited in two directions: a) to provide fast search procedures in VQ; and b) to obtain an efficient encoding of the LPC envelope. These two topics are discussed under a proposed 2400 bps CELP coder, that shows good quality in the reproduced speech.

Full Paper

Bibliographic reference.  Lopez-Gonzalo, Eduardo / Hernandez-Gomez, Luis A. (1993): "Fast vector quantization using neural maps for CELP at 2400bps", In EUROSPEECH'93, 55-58.