摘要 |
PURPOSE: A device and a method for coding a voice are provided to enhance quality of voice spectrum approximating for both voice and voiceless band. CONSTITUTION: A device for coding a voice contains a voice model parameter decider(1) and a parameter encoder(2). The voice model parameter decider comprises a rectangular windowing unit(10), a hamming window unit(20), a frame classification unit(30), a fast Fourier transform unit(40), a pitch detector(3), a V/UV discrimination unit(80) and a spectrum size decider(90). With the components, the device decides a MBE model parameter such as pitch frequency, planetary deciding group, a spectrum size group and a phase value.
|