To provide a voice-encoding system which can output encoded data, having a different sampling frequency and the different number of channels from an input audio signal, without excessively increasing the processing quantity needed for encoding processing using a simple constitution.

A subband sample group Si, converted to a frequency area which, is originally generated in the encoding processing is used for a time area signal as an input signal, successive sub-band sample group Si data are thinned out or interpolated and encoded by using the data, so that the encoded data which has a different sampling frequency and a different number of channels from the input audio signal (different from the input form) can be outputted.