| 摘要 |
PURPOSE:To make the compression-encoding time of a voice encoding device shorter by automatically stopping the compression-encoding of voice signals at the moment when the degree of quantization error exceeds a prescribed value and restarting the compression-encoding by increasing the number of quantization bits. CONSTITUTION:An absolute value device 16 selects either one of a quantization error signal and the output signal of a noise shaping filter 12 to which the error signal is inputted and obtains its absolute value and inputs the absolute value to a maximum value detector 17a and cumulative adder 18a. Then the device 16 detects the maximum value and cumulative sum of the quantization errors produced in the course of compression-encoding of inputted sound signals as statistical quantities. When the maximum value and cumulative sum exceed prescribed values, the device 16 discriminates that the quality of the sounds deteriorates and stops the compression- encoding, and then, automatically increases the number of quantization bits and causes the compression-encoding to be restarted from the beginning. Therefore, the time required for making the compression-encoding can be reduced sharply and, when the compression-encoding is once started, no man is required until the completion of the compression-encoding. |
