Prosody Generation in a Chinese TTS System Based on a Hierarchical Word Prosody Template Tree

In this paper, a prosody generation method based on a hierarchical word prosody template tree for the generation of prosodic information in a sentence is proposed. The hierarchical word prosody template tree is established from a large continuous speech database according to the linguistic features: tone combination, word length, part of speech (POS) of the word, and word position in a sentence. This template tree stores the prosodic features including pitch contour, energy contour, and syllable duration of a word for possible combinations of linguistic features. In the inside test, the prosody of the synthesized speech resembles that of the original speech for a typical sentence.