PURPOSE: To provide possibility of decoding voices so that the background noise does not becomes unnatural by accumulating parameters of the background noise fed intermittently, and making interpolation between the parameter received anew and the parameter accumulated.
CONSTITUTION: A symbol string obtained by encoding the voice spectrum factor of voice signal and the component(s) which can not be represented by this factor, is fed to an input end 1, and an energize signal producing circuit 2 and a voice spectrum factor producing circuit 3 produce an energize signal and a voice spectrum factor, respectively, from the fed symbol string. From this output, a synthesizing filter 6 reproduces a voice signal, while a voice output circuit 7 emits the background noise continuously until a new voice signal is reproduced with the next symbol in case symbol strings are fed intermittently because of the voice signal being voiceless. At this time, parameters of background noise fed intermittently are accumulated in a voice spectrum factor holding buffer 4, and a voice spectrum factor interpolating circuit 5 make interpolation between the parameter received anew and the parameter accumulated.
UNNO YOSHIHIRO
JPH02282798A | 1990-11-20 | |||
JPH05122165A | 1993-05-18 | |||
JPS62102300A | 1987-05-12 | |||
JPS6123200A | 1986-01-31 |