15th International Congress of Phonetic Sciences (ICPhS-15)

Barcelona, Spain
August 3-9, 2003


Relationship Between Control Precision and Perceptual Sensitivity to Segmental Duration

Hiroaki Kato (1), Makiko Muto (2), Minoru Tsuzaki (3), Yoshinori Sagisaka (2)

(1) ATR-HIS, Japan
(2) Waseda University, Japan
(3) ATR-SLT, Japan

This paper investigates a potential linkage between production and perception in the temporal aspect of speech from an ecological point of view. The variability of the segment duration is analyzed to estimate the temporal control precision in speech production, using large-scale speech corpora comprising a commonly used word set (5240 word entries spoken by 20 speakers) and a phonetically balanced sentence set (503 sentence entries spoken by 10 speakers). The results first show that the duration of a given segment has little correlation with the estimated control precision of that segment. They also show that the estimated precision systematically changes as a function of vowel quality and temporal position in an utterance unit, such as a word or phrase. These findings are in good agreement with the perceptual characteristics, i.e., sensitivity and acceptability of deviation in segment duration; accordingly, the results suggest the existence of some linkage between perception and production.

Full Paper

Bibliographic reference.  Kato, Hiroaki / Muto, Makiko / Tsuzaki, Minoru / Sagisaka, Yoshinori (2003): "Relationship between control precision and perceptual sensitivity to segmental duration", In ICPhS-15, 2043-2046.