PURPOSE: To provide a voice having high natural properties by performing continuous compression of a frequency spectrum in a way that a band of BPF at a receiving end part for a reproduction sound of a specified times speed forms a band provided by multiplying the band of a BPF at a feed end part by a specified times inverse number.
CONSTITUTION: The frequency of a voice reproduced at a double speed by a tape recorder 1 is analyzed by means of band filters 31 - 316, is detected by means of rectifiers 81 - 816, and fetched as spectral comprehensive information through BPF 91 - 916 to feed the information to the receiving end side. Voice source information obtained by a voice and silence detector 5 and a pitch sampling device 6 is fed to a receiving end part. The receiving end part selects a voice and silence by means of a pulse voice generator 10, a noise generator 11, and a switch 12 based on voice source information. In the case of a voice being present, the pitch frequency of a pulse from the generator 10 is decided and fed as a reproduction voice source to a reproduction voice source to modulators 131 - 1316, and a modulating output is outputted to a speaker 14. Since, at the receiving end part, the band of filters 41 - 416 is 1/2 of that of filters 31 - 316, and a frequency reproduced and spread to a double value is compressed to 1/2 to produce an original frequency.
YAMAMURA AKIRA
OMUKAI JUNKO
Next Patent: VOICE DATA PRODUCING SYSTEM FOR VOICE SYNTHESIZING DEVICE