Title :
Syllable clustering and spectral discontinuity in syllable-based TTS systems
Abstract :
The paper examines the spectral discontinuity problem existing in syllable-based Chinese TTS (text-to-speech) systems. Acoustic and phonetic investigations showed that, in natural speech, syllables with approximant, nasal or vowel as onset have a tendency to form syllable clusters with their preceding syllables due to the strong coarticulation effect. In speech synthesis, syllable clusters are the major source for audible spectral discontinuity. The implication of this finding for improving syllable-based TTS voice quality is discussed.
Keywords :
natural languages; pattern clustering; speech; speech synthesis; Chinese TTS systems; acoustic investigations; coarticulation effect; natural speech; phonetic investigations; spectral discontinuity; speech synthesis; syllable clustering; text-to-speech systems; voice quality; Acoustic noise; Clustering algorithms; Energy states; Laboratories; Loudspeakers; Natural languages; Spectrogram; Speech synthesis;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). 2003 IEEE International Conference on
Print_ISBN :
0-7803-7663-3
DOI :
10.1109/ICASSP.2003.1198874
