Third International Conference on Spoken Language Processing (ICSLP 94)

Yokohama, Japan
September 18-22, 1994

Prosodic Pattern of Utterance Units in Japanese Spoken Dialogs

Kazuyuki Takagi, Shuichi Itahashi

Institute of Information Sciences and Electronics, University of Tsukuba, Tsukuba, Ibaraki, Japan

This paper reports a study to classify spontaneously spoken utterances by their prosodic features only. First, input speech was segmented into utterance units, each of which is a speech interval segmented by pauses or distinctive utterances such as interjections and fillers. Secondly, the cluster analysis was conducted using seven prosodic parameters of an utterance unit, i.e., speech rate, F0 onset, final, values and the positions of the maximum and the minimum Fq. On the 10 spoken dialogs uttered by 9 speakers selected from ASJ Continuous Speech Corpus, the utterance units were classified into 5 distinctive clusters: "Initial & Middle", "Final", "Slow", "Devoiced", and "Others". Each cluster was characterized by syntactic and prosodic features of utterance units contained in the cluster. Results of this paper can be used for a preprocessing of spontaneous speech understanding systems.

Full Paper

Bibliographic reference.  Takagi, Kazuyuki / Itahashi, Shuichi (1994): "Prosodic pattern of utterance units in Japanese spoken dialogs", In ICSLP-1994, 143-146.