14th International Congress of Phonetic Sciences (ICPhS-14)

San Francisco, CA, USA
August 1-7, 1999


Joint Estimation of Voice Source and Vocal Tract Parameters as Applied to the Study of Voice Source Dynamics

Hideki Kasuya (1), Kikuo Maekawa (2), Shigeru Kiritani (3)

(1) Faculty of Engineering, Utsunomiya University, Utsunomiya, Japan
(2) The National Language Research Institute, Tokyo, Japan
(3) Graduate School of Medicine, University of Tokyo, Tokyo, Japan

A novel method is presented for the joint estimation of voice source and vocal tract (formant/anti-formant) parameters from the acoustic speech signal. The method is based on the ARX (auto-regressive with exogenous input) model with a glottal flow waveform as an input to the system. A nonlinear optimization strategy is employed to estimate glottal flow parameters, whereas an extended Kalman filter algorithm is used to extract vocal tract parameters. The amount of turbulence noise components included in the voiced speech segment is also measured by a bank of band pass filters. The method is applied to the study of voice source dynamics as a conveyer of paralinguistic and vocal quality features that include perceived naturalness of an utterance. It is shown that this method is useful to gain profound understanding of voice source dynamics when used under certain limitations.

Full Paper

Bibliographic reference.  Kasuya, Hideki / Maekawa, Kikuo / Kiritani, Shigeru (1999): "Joint estimation of voice source and vocal tract parameters as applied to the study of voice source dynamics", In ICPhS-14, 2505-2512.