Abstract
A method to extract excitation signals for voice synthesis system is presented. Since the human voice does not have perfect periodicity, high-quality voice synthesis systems generally use the mixed excitation signal to represent the aperiodic signal. In STRAIGHT and TANDEM-STRAIGHT, the aperiodicity is defined as the spectrum to synthesize both periodic and aperiodic signals. The proposed system is the same as TD-PSOLA, provided that voice is synthesized without spectral modification. On the other hand, it can modify the spectral envelope and synthesize natural voice. These results suggest that the proposed system has advantages over both STRAIGHT and TD-PSOLA. Voice synthesized with the proposed system is superior to the conventional systems in sound quality.
Original language | English |
---|---|
Pages (from-to) | 123-125 |
Number of pages | 3 |
Journal | Acoustical Science and Technology |
Volume | 33 |
Issue number | 2 |
DOIs | |
Publication status | Published - 2012 |
Keywords
- Excitation signal
- Vocoder
- Voice analysis
- Voice synthesis