한국음향학회:학술대회논문집 (Proceedings of the Acoustical Society of Korea Conference)
- 한국음향학회 1994년도 FIFTH WESTERN PACIFIC REGIONAL ACOUSTICS CONFERENCE SEOUL KOREA
- /
- Pages.984-990
- /
- 1994
SPEECH SYNTHESIS IN THE TIME DOMAIN BY PITCH CONTROL USING LAGRANGE INTERPOLATION(TD-PCULI)
- Kang, Chan-Hee (Dept. of Electronics in Sangji junior college) ;
- Shin, Yong-Jo (Dept. of Electronics in Sangji junior colleg) ;
- Kim, Yun-Seok- (Dept. of Electronics in Sangji junior colleg) ;
- Kang, Dae-Soo (Telesys ERC Telemation System Engineering Research Cente) ;
- Lee, Jong-Heon- (Telesys ERC Telemation System Engineering Research Cente) ;
- Kwon, Ki-Hyung (Dept of Electronic Engineering in Kyunghee Uni) ;
- An, Jeong-Keun (Dept of Electronic Engineering in Kyunghee Uni) ;
- Sea, Sung-Tae (Dept of Electronic Engineering in Kyunghee Uni) ;
- Chin, Yong-Ohk (Dept of Electronic Engineering in Kyunghee Univ)
- 발행 : 1994.06.01
초록
In this paper a new speech synthesis method in the time domain using mono-syllables is proposed. It is to overcome the degradation of the synthetic speech quality by the synthesis method in the frequency domain and to develop an algorithm in the time domain for the prosodic control. In particular when we use a method in a time domain with mono-syllable as a synthesis unit it will be the main issues which are to control th pitch period and to smooth the energy pattern. As a solution to the pitch control, a method using Lagrange interpolation is suggested. As a solution to the other problem, an algorithm which can control the amplitude envelop shape of mono-syllable is proposed. As the results of experiments it was possible to synthesize unlimited Korean speeches including the prosody control. Accoding to the MOS evaluation the quality and the naturality in them was improved to be a good level.
키워드