Sampled sound data is compressed with a vector quantizing technique and
then transmitted via a communication line. Received sound data is decoded,
compressed with an ADPCM technique, and then stored into a memory. In
response to a request for reproduction, the ADPCM sound data is read out,
decoded, and then sounded. As another example, in a karaoke device, sample
sound data is supplied after being compressed with the vector quantizing
technique, in addition to MIDI-form music performance data. A music sound
is reproduced on the basis of the MIDI-form music performance data, and at
the same time a sound is reproduced by decoding the vector-quantized sound
data. As another example, in a karaoke device, data obtained by
compressing sampled sound data with the vector quantizing technique is
mixed with data obtained by compressing sampled data with the ADPCM
technique, and in reproduction, a predetermined decoding process is
executed after identifying with which of technique is compressed data to
be reproduced. As still another example, in a game device, sampled sound
data, of human voice, effect sound, etc. are prestored after being
compressed with the vector quantizing technique, so that in accordance
with progression of a game, the data are read out and decoded for
reproductive sounding.