Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 95 === In this thesis, the main search goal is to develop a continuous speech time and pitch scaling algorithm. By using vocoder model and spectral excitation, which are used in speech coding techniques, we try to find the corresponding parameters to achieve effective speech time and pitch scaling technique. By using the melody information of MIDI signal, we can adjust pitch and rhythm of the singer by the developed speech time and pitch scaling techniques. Finally, the modification of pitch and time of the song, we complete an intelligent karaoke system, which can automatically correct the erroneous pitch and rhythm of the singer
|