Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Adaptive quantizing in speech coding systems

: Noll, P.

Institute of Electrical and Electronics Engineers -IEEE-:
Source encoding, channel encoding and modulation, digital communication loops : Digital communications, International Zurich Seminar. Proceedings. 12 - 15 mar 1974
New York, NY: IEEE, 1974
International Seminar "Digital Communications" <1974, Zürich>
Fraunhofer HHI ()
digital communication systems; encoding; pulse-code modulation; speech; voice communication; adaptive quantizing; speech coding systems; local estimate; time dependent variance; digital speech encoding system; digital communication system

This paper deals with the problem of calculating a local estimate of the time dependent variance of speech signals. The estimation value controls an adaptive amplifier, placed in front of a quantizer, to maintain the quantizer input signal within a specified range. A forward and some backward estimation schemes have been studied and optimized. Results are given for various optimum 3 bit-quantizers. A nonadaptive 3 bit-quantizer has a signal-to-noise ratio (SNR) of approximately 8-9 dB; an adaptive scheme not only raises this value up to 14-15 dB, but leads to an additional subjective gain. DPCM systems offer an additional rise of SNR up to 19 dB.