摘要 |
PURPOSE:To prevent the head of speech from being cut off and the quality of voice from being deteriorated by the constitution that a minimum value of a residual difference quantized step is limited in a voice adaptive forecast coding circuit using entropy coding. CONSTITUTION:A residual quantized step (d) is limited to a dmin at a novoice section or a minute voice section where the power of a signal (x) is small by allowing a residual quantizing step limit means 30 to set the minimum value dmin of the residual quantized step to a large value when the bit number N2 stored in a data storage means 22 is large. Thus, the number of bits N1 is small so as to multiplex lots of data y' (N-N1-bit). |