JAIST Repository >
School of Information Science >
Conference Papers >
Conference Papers >

Please use this identifier to cite or link to this item: http://hdl.handle.net/10119/15770

Title: Study on Relationship between Degree of Emphasis and Acoustic Feature for Synthesizing Emphasized Speech
Authors: Ohtani, Yasuhiro
Akagi, Masato
Issue Date: 2019-03-06
Publisher: Research Institute of Signal Processing, Japan
Magazine name: 2019 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP2019)
Start page: 256
End page: 259
Abstract: Humans can perceive not only presence/absence of emphasis but also degrees of emphasis from actual emphasized speech. However, humans cannot fully do from synthesized speech. This paper focused on two properties of Fundamental frequency (F0) contours: amount of decay from the accent nucleus and variation between each accent nucleus, and hypothesized that the two properties of F0 contours are important for synthesizing emphasized speech. To discuss this hypothesis, this paper clarified relationships between degrees of emphasis and F0 contours. To clarify relationships, it was necessary to compare relationships for each stimulus. To compare relationships, it was necessary to know the degree of emphasis of each stimulus and analyze variations of F0 contours. A listening test was carried out to obtain the degrees of emphasis of stimulus. A value which is frequency at the barycentric point of the vowel was extracted from F0 contours to analyze the variation of F0 contour. From these results, we had two findings; degree of emphasis is increasing when amount of decay from accent nucleus to next mora is increasing, and the variation of accent nuclei is different with/without emphasis. The experiment was carried out to evaluate hypothesis. Synthesized stimuli from non-emphasized voice by varying amount of decay and variation of accent nuclei are used for the experiment. The results showed that the participants of the experiment can perceive emphasis with degrees from the synthesized stimuli. This result clarified the relationships between presence/absence of emphasis and two findings. In addition the hypothesis is important for synthesizing emphasized speech which convey presence/absence of emphasis.
Rights: Copyright (C) 2019 Research Institute of Signal Processing, Japan. Yasuhiro Ohtani and Masato Akagi, 2019 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP2019), 2019, 256-259.
URI: http://hdl.handle.net/10119/15770
Material Type: publisher
Appears in Collections:b11-1. 会議発表論文・発表資料 (Conference Papers)

Files in This Item:

File Description SizeFormat
2921.pdf1140KbAdobe PDFView/Open

All items in DSpace are protected by copyright, with all rights reserved.

 


Contact : Library Information Section, Japan Advanced Institute of Science and Technology