Research Of Grey Algorithm Of Low Rate Speech Compression Based On DSP

The voice codec is developing quickly since 1930s', through 70 years' research, up to 1990, the technology of computer and the communication push the voice codec, especially lately in 20^th century, the computer technique, and the massive IC circuit with the rapid development of modern communication.The low rate voice coding is the main technology in voice codec. In modern communication, the use of channel resource is important. Now that the channel is difficult to expand and the quality is bad, the low rate voice coding is essential, with the development of telecom, the research of the low rate voice coding of high quality is urgent. It is well used in the field like net communication, secret communication, voice mail, IP phone, especially in information troops. ADPCM uses discrete time to sample signal from changing analog signal by sampling amplitude, to be the extent of PAM. PAM system can be changed into PCM system through adding AD to signal source. Typical quantized process is single quantized value, to introduce some error in signal sampling. The error is so-called quantized noise. It can be minimized by reducing sampling interval. As quantize increased, the bits are increased simultaneously.The paper describes a new algorism based on grey system that system is also applied to voice field. In the processing of data, the grey theory collects the regularity of the original data, and the other ways of processing theory use the static and experience based on many data. But grey has no limit, the grey theory needs only above four data to set up a model, not needing any data's feature. To use the model GM(1,1) to build nonlinear predictor, accumulate the data to get less data, and to code the data by ADPCM codec to get lower rate voice coding.