Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Roles of Voice Source Dynamics As a Conveyer of Paralinguistic Features

Hideki Kasuya (1), Masanori Yoshizawa (1), Kikuo Maekawa (2)

(1) Faculty of Engineering, Utsunomiya University, Japan
(2) National Language Research Institute, Japan

This paper focuses on the role of frequency spectral tilt (TL) of the voicing source waveform, the glottal turbulent noise and fluctuations of the source spectrum on period-to-period basis in the cognition of paralinguistic features as in suspicious and disappointed renditions. The TL is closely related to duration of the return phase of the differentiated glottal waveform. Perceptual experiments have shown that the larger the TL, the lower the degree of suspicion and existence of the source spectral fluctuations increases that degree over the entire range of the TL. In the case of disappointment, the larger the TL, the higher the degree of disappointment and inclusion of the glottal turbulent noise emphasizes disappointment when the TL is small.


Full Paper

Bibliographic reference.  Kasuya, Hideki / Yoshizawa, Masanori / Maekawa, Kikuo (2000): "Roles of voice source dynamics as a conveyer of paralinguistic features", In ICSLP-2000, vol.2, 345-348.