Title :
Optimization of intonation control using statistical F0 resetting characteristics
Author :
Sagisaka, Yoshinori ; Kaiki, Nobuyoshi
Author_Institution :
ATR Interpreting Technology Res. Lab., Kyoto, Japan
Abstract :
F0 resetting characteristics were analyzed statistically in relation to local phrase dependency structure. Analyses showed that the parameters reflecting the local phrase structure and coordinate phrase relation could explain F0 resetting characteristics quantitatively. An F0 prediction experiment was carried out using four models corresponding to four different types of local structure. The F0 prediction results showed that the total prediction error was reduced 20% (about 0.5 semitone) by the differentiation of local phrase structure and the distinction between coordinate and noncoordinate phrase boundaries
Keywords :
optimisation; speech analysis and processing; speech synthesis; statistical analysis; F0 prediction experiment; coordinate phrase boundaries; coordinate phrase relation; fundamental frequency; intonation control; local phrase dependency structure; models; noncoordinate phrase boundaries; optimisation; speech; statistical F0 resetting characteristics; Accuracy; Data analysis; Databases; Laboratories; Predictive models; Shape; Speech analysis; Speech synthesis; Statistical analysis; Telephony;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0532-9
DOI :
10.1109/ICASSP.1992.226123
