Synthesizing intonation for computer speech output
As a basis for designing an improved algorithm for synthetic intonation, two existing algorithms were implemented and tested by comparison with natural speech. The problems are discussed and the improved implementation is described. Interestingly, the simple approach by Witten was found to be as natural as the more complex approach by Collier and Hart and their co-workers within the constraints of this thesis, and was chosen as the basis for and improved algorithm. The results exceeded expectations but further rounds of experimentational evaluation and improvements are necessary.
Bibliography: p. 72-74.
Taube-Schock, C. (1994). Synthesizing intonation for computer speech output (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/13649