Presentation 2004/8/12
Pause prediction from input Chinese text by analyzing part-of-speech sequences
Jinfu Ni, Hisashi KAWAI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper presents a practical method for predicting pause positions in text-to-speech synthesis. An input sentence is analyzed into phrase chunks based on both punctuation and a pattern of part-of-speech (POS) tags of underlying words that matches a pre-selected POS pattern. A set of pre-selected POS patterns represents possible phrasal structures that are extracted from restricted syntactic clauses in an unsupervised way. Maximum likelihood estimation and rules-based techniques are jointly used to decode POS patterns from the sentence for phrasing. Simulation results indicate that 83.7% of predicting pause positions are consistent with those measured in real speech, with 64.2 % accuracy in the case of a pause without accompanying punctuation.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Pause prediction / Morphological analysis / Part of speech / Text-to-speech
Paper # SP2004-47
Date of Issue

Conference Information
Committee SP
Conference Date 2004/8/12(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Speech (SP)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Pause prediction from input Chinese text by analyzing part-of-speech sequences
Sub Title (in English)
Keyword(1) Pause prediction
Keyword(2) Morphological analysis
Keyword(3) Part of speech
Keyword(4) Text-to-speech
1st Author's Name Jinfu Ni
1st Author's Affiliation ATR Spoken Language Translation Research Laboratories.()
2nd Author's Name Hisashi KAWAI
2nd Author's Affiliation ATR Spoken Language Translation Research Laboratories.
Date 2004/8/12
Paper # SP2004-47
Volume (vol) vol.104
Number (no) 252
Page pp.pp.-
#Pages 6
Date of Issue