Speech Prosody 2004

Nara, Japan
March 23-26, 2004

Study on Pitch Contour of Thai Polysyllabic Tone Sequences Using a Generative Model

Pusadee Seresangtakul (1), Tomio Takara (2)

(1) Department of Computer Science, Khon Kaen University, Khon Kaen, Thailand
(2) Department of Information Engineering, University of the Ryukyus, Okinawa, Japan

Thai speech synthesis by rule has been developed. In order to synthesize F0 contours of Thai tones, the generative model of F0 contours (Fujisaki's model) for tonal languages is applied. Along with our method, the pitch contours of Thai polysyllabic words were analyzed. Rules are derived and applied to synthesize Thai polysyllabic tone sequences. We performed listening tests to evaluate intelligibility of the model for Thai tone generation. The average intelligibility scores were 98.8%and 96.6% for disyllabic and trisyllabic words, respectively. The generative model of F0 contours for Thai words was shown to be effective. Furthermore, we derived rules to synthesize suprasegmental F0 contours using the trisyllabic words' parameters. We performed listening tests to evaluate the intelligibility score and naturalness of synthesized speech. As a result, all phrases/sentences were completely identified. The MOSs (Mean Opinion Score) was 3.50 while the original and analysis/synthesis samples were 4.82 and 3.59, respectively.

Full Paper

Bibliographic reference.  Seresangtakul, Pusadee / Takara, Tomio (2004): "Study on pitch contour of Thai polysyllabic tone sequences using a generative model", In SP-2004, 483-486.