EUROSPEECH 2001 Scandinavia
7th European Conference on Speech Communication and Technology

Aalborg, Denmark
September 3-7, 2001


Systematic F0 Glitches around Nasal-Vowel Transitions

Hideki Kawahara (1), Parham Zolfaghari (2)

(1) Wakayama University / ATR-ISD / CREST, Japan
(2) CIAIR, Nagoya University, Japan

High-resolution F0 analysis using a speech database with simultaneously recorded EGG (Electroglottogram) signals indicated that there are systematic F0 glitches around nasal-vowel transitions. The durations of the glitches are 10 to 20 ms and they introduce 5 to 10 Hz F0 shifts. A detailed series of analyses of these glitches indicated that the major contributing factor of these glitches is sudden changes of group delay values of the vocal tract transfer function in the vicinity of the fundamental frequency at nasal-vowel transitions. It is also suggested that the Doppler effects due to apparent changes of vocal tract length are marginal, even if they exist. Finally, issues in evaluating high resolution F0 extraction algorithms and applications to high quality speech manipulation methods are discussed.

Full Paper

Bibliographic reference.  Kawahara, Hideki / Zolfaghari, Parham (2001): "Systematic F0 glitches around nasal-vowel transitions", In EUROSPEECH-2001, 2459-2462.