JPS5774809A - Voice signal processing device - Google Patents

Abstract

PURPOSE:To obtain a synthesized sound being almost same as its original sound, by dividing a voice signal to a time length segment being almost equal to integer times of a basic period, and procesing it, when a voice signal recorded in a recording medium such as a magnetic tape, etc. is reproduced at a speed which is different from that of a recording time. CONSTITUTION:A reproducing signal from a tape recorder is inputted from a terminal 1, is A/D converted 2 and is stored in buffers 3, 4. The buffers 3, 4 are constituted of a shift register or an RAM, are read out alternately through a changeover switch 5, and after that, are D/A converted 6 and are outputted from a terminal 7. On the other hand, the output which has been A/D converted 2 is supplied to a pitch extractor 8, and the extracted pitch synchronization is supplied to a buffer controller 9. The controller 9 controls not only write-in and read-out of the buffers 3, 4 but also a clock, and changes over the switch 5, in accordance with the extracted synchronization. According to such a constitution, the pitch is arrayed repeatedly in case of a low speed reproduction and is arrayed by being thinned out in case of a high speed reproduction, and a reproducing sound being almost same as its original sound is obtained.