摘要 |
PURPOSE:To facilitate addressing in voice reproduction and enable natural voice reproduction by compressing no-voice sections of a voice signal into frames which have constant length regardless of the length of the no-voice sections. CONSTITUTION:A no-voice detection part 1 detects the no-voice parts from the voice signal S1 and a no-voice counter 2 counts the number of the detected frames. A voice encoding part 4 encodes the voice to be encoded in the voice signal S1. Namely, the voice encoding part 4 encodes the voice parts as they are and compresses the no-voice parts to output data together with the counted value of the no-voice counter 2, so that they are recorded in a memory 5 for code storage. When it is judged that the no-voice part is longer than the specific memory 5 for code storage, and an address generator 3 stops advancing the address from the point of time. Then when the voice part of the voice signal S1 is inputted, the advance of the address is restarted and the encoded voice signal is stored in the memory 5. |