| 摘要 |
The lower significant n bits of a first differential value between adjacent sampled values of a sampled received signal are cut off to provide a second differential value, which is then subjected to adaptive quantization to thereby provide data compression. When this data is expanded, the second differential value decoded from the adaptive quantized value is sequentially accumulated 2n times at a period corresponding to 1/2of the sampling period of the received signal. As a result, a quantizing error (quantization noise) per sampling of the received signal is dispersed in 2n periods to thereby provide an improved acoustic S/N ratio. When the first differential value of the received signal is quantized, when the lower significant n bits of the first differential value are cut off, and when the resulting signal is subjected to adaptive quantization, the resulting respective quantizing errors are fed to corresponding values processed next. Thus, accumulation of the quantizing errors is prevented.
|
