An 800 Bps real-time voice coding system based on efficient encoding techniques
暂无分享,去创建一个
A real-time 800 bit per second fixed-rate voice communication system has been developed and implemented on the TRW Microprogrammable Voice Processor (MVP) using 11K words of memory. DRT and other test results indicate acceptable intelligibility and communicability. This low data rate is achieved by an algorithm which efficiently encodes the parameters of 2400 bps LPC. This paper describes the essential components of the algorithm along with details of implementation and test results. Rather than attempt formant tracking, as many previous approaches to this data rate have done, this new algorithm applies a number of data reduction schemes, common in other types of communication systems, to achieve the desired data rate. These schemes include adaptive DPCM, Karhunen-Loeve transformation, statistically optimized quantization, and dynamic bit allocations.
[1] Joel Max,et al. Quantizing for minimum distortion , 1960, IRE Trans. Inf. Theory.
[2] F. Itakura,et al. Minimum prediction residual principle applied to speech recognition , 1975 .
[3] Thomas P. Barnwell,et al. An analysis of objective measures for user acceptance of voice communications systems , 1979 .
[4] R. Gray,et al. Speech coding based upon vector quantization , 1980, ICASSP.