Correct prosodic boundary prediction is crucial for the quality of synthesized speech in text-to-speech system. This article mainly presents the prosodic hierarchy of Uyghur language, which belongs to Turkish language...Correct prosodic boundary prediction is crucial for the quality of synthesized speech in text-to-speech system. This article mainly presents the prosodic hierarchy of Uyghur language, which belongs to Turkish language family of Altaic language system and further verifies the reliability of proposed Uyghur prosodic boundary annotation rules by acoustic analysis. In the prediction part, a two-layer shifting hierarchical approach based on decision tree is used for predicting prosodic word and prosodic phrase boundary, and the influence of different feature sets on the Uyghur prosodic boundary prediction is also investigated. Experimental results clearly show the acoustical changes and automatic prediction performance of different prosodic boundaries of Uyghur language, thus laying a good foundation for further research.展开更多
According to Register Grammar,prosody,as an aspect of grammar,is one way to realize different registers.This study explored the differences in the acoustic features of prosodic boundaries between Chinese formal and in...According to Register Grammar,prosody,as an aspect of grammar,is one way to realize different registers.This study explored the differences in the acoustic features of prosodic boundaries between Chinese formal and informal speech.Results suggested that:(1) Pauses occurred more frequently and lasted longer at prosodic boundaries in formal speech,best reflected at the Prosodic Clitic level and at the Prosodic Phrase level respectively.In formal speech,pauses at Prosodic Phrase boundaries lasted significantly longer than those at Prosodic Clitic boundaries,while this difference was not significant in informal speech.The distribution of pause duration displayed greater dispersion as the prosodic level increased.(2) In informal register,Prosodic Phrase boundaries performed higher degrees of pre-lengthening than Prosodic Clitic boundaries,while this difference was not significant in formal speech.Prosodic Clitic boundaries in formal and informal speech displayed pre-lengthening and postlengthening,respectively.(3) Pre-strengthening in the intensity of prosodic words at prosodic boundaries existed at all three levels in both registers,but it was probably a weak cue to discriminate the two registers.(4) Only slight pitch reset was found at Prosodic Clitic boundaries in formal speech and at Prosodic Phrase boundaries in informal speech.展开更多
基金Supported by the National Natural Science Foundation of China(61065005and61062008)
文摘Correct prosodic boundary prediction is crucial for the quality of synthesized speech in text-to-speech system. This article mainly presents the prosodic hierarchy of Uyghur language, which belongs to Turkish language family of Altaic language system and further verifies the reliability of proposed Uyghur prosodic boundary annotation rules by acoustic analysis. In the prediction part, a two-layer shifting hierarchical approach based on decision tree is used for predicting prosodic word and prosodic phrase boundary, and the influence of different feature sets on the Uyghur prosodic boundary prediction is also investigated. Experimental results clearly show the acoustical changes and automatic prediction performance of different prosodic boundaries of Uyghur language, thus laying a good foundation for further research.
基金supported by Social Science Foundation of Tianjin,China (TJWW19-009 and TJWW17-010)
文摘According to Register Grammar,prosody,as an aspect of grammar,is one way to realize different registers.This study explored the differences in the acoustic features of prosodic boundaries between Chinese formal and informal speech.Results suggested that:(1) Pauses occurred more frequently and lasted longer at prosodic boundaries in formal speech,best reflected at the Prosodic Clitic level and at the Prosodic Phrase level respectively.In formal speech,pauses at Prosodic Phrase boundaries lasted significantly longer than those at Prosodic Clitic boundaries,while this difference was not significant in informal speech.The distribution of pause duration displayed greater dispersion as the prosodic level increased.(2) In informal register,Prosodic Phrase boundaries performed higher degrees of pre-lengthening than Prosodic Clitic boundaries,while this difference was not significant in formal speech.Prosodic Clitic boundaries in formal and informal speech displayed pre-lengthening and postlengthening,respectively.(3) Pre-strengthening in the intensity of prosodic words at prosodic boundaries existed at all three levels in both registers,but it was probably a weak cue to discriminate the two registers.(4) Only slight pitch reset was found at Prosodic Clitic boundaries in formal speech and at Prosodic Phrase boundaries in informal speech.
