Commonalities of Glottal Sources and Vocal Tract Shapes Among Speakers in Emotional Speech

Discussion on commonalities of the glottal source waves and vocal tract shapes among speakers in emotional-speech production can help us to carry out further speech processing from the speech-production point of view. This requires accurate and independent measurement of glottal source waves and vocal tract shapes. Although magnetic resonance imaging and electromagnetic articulography can help achieve this requires, they are time consuming and costly. This paper explores the commonalities of the glottal source waves and vocal tract shapes among speakers in emotional speech based on a source-filter model. With the help of the proposed precise estimation algorithm, the glottal source waves and vocal tract shapes of emotional voices (vowel:/a/) were estimated, in which the emotional speech with four basic emotional speaking styles (neutral, joy, anger and sad) were uttered by four different speakers. The results are as follows. When compared with the spectra tilts of glottal source waves of the neutral emotion, (1) those of anger and joy increased, and that of sad decreased in the 200- to 700-Hz frequency range; (2) that of anger increased, but that of joy decreased, and that of sad was the same as that of neutral in 700- to 2000-Hz range; (3) all spectral tilts had the same tendency over 2000 Hz. For front vocal tract shapes, the area function of anger was the largest, that of sad was smallest, and those of joy and neutral were in the middle.11th International Seminar, ISSP 2017, Tianjin, China, October 16-19, 2017Lecture Notes in Computer Science (LNCS) vol.10733 - Studies on Speech Productio