First International Conference on Spoken Language Processing (ICSLP 90)

Kobe, Japan
November 18-22, 1990

A Support Environment Based on Rule Interpreter for Synthesis by Rule

Yoichi Yamashita (1), Hiroyuki Fujiwara (2), Yasuo Nomura (2), Nobuyoshi Kaiki (3), Riichiro Mizoguchi (1)

(1) I.S.I.R., Osaka University, Ibaraki-shi, Osaka, Japan
(2) Faculty of Engineering, Kansai University, Suita-shi, Osaka, Japan
(3) Sharp Corporation, Tenri-shi, Nara, Japan

SSRI (Speech Synthesis system using Rule Interpreter) is a support environment for the experts to develop rule bases of speech synthesis. In SSRI, the synthesis rules are stored in SSRB (Speech Synthesis Rule Base) and separated from RI (Rule Interpreter) which generates the parameter transition contours. In general, management of the fired rules for each text can reduce the time of speech re-synthesis from the text which has ever converted to speech by RI. RM (Rule Manager) keeps the history of rule invocation and restores it after the modification of SSRB. And, RM divides the synthesis rules into groups to accelerate speech synthesis from a new text. The history data and the rule group drastically reduce the turn-around time of synthesis rule development. SSRI also provides the user with several utilities of graphically displaying the parameter transition contours, listing the fired rules for the text, saving the synthesized speech wave and so on. SSRI is implemented in C-Prolog and C on the UNIX engineering workstation.

Full Paper

Bibliographic reference.  Yamashita, Yoichi / Fujiwara, Hiroyuki / Nomura, Yasuo / Kaiki, Nobuyoshi / Mizoguchi, Riichiro (1990): "A support environment based on rule interpreter for synthesis by rule", In ICSLP-1990, 769-772.