PURPOSE:To obtain a comparatively large number of vocabularies and a large number of simultaneous response channels without increasing the memory capacity of a storage medium by compressing the information volume of sound waveforms. CONSTITUTION:Word sound waveforms are inputted to analyzing part 102 and have the pitch and the amplitude width extracted and are stored successively in composition data memory 103. Composition data distribution circuit 104 transfers sound one long time frame components of composition data from memory 103 to buffer memories 106, 107 and 108 (double buffer system) of respective channels. Micro processors 110, 111 and 112 use composition data, which are stored in buffer memories 106, 107 and 108, to edit and compose phoneme piece waveforms and outputs digital composed waveform data to DA converters 113, 114 and 115. At this time, microprocessors use data in buffer memory A when status flag 109 is ''1'', and use data in B when flag 109 is ''0''.
展开▼