ASA 127th Meeting M.I.T. 1994 June 6-10

1pSP41. X-ray film database for speech research.

K. G. Munhall

Dept. of Psychol., Queen's Univ., Kingston, ON K7L 3N6, Canada

E. Vatikiotis-Bateson

Y. Tohkura

ATR Human Inf. Process. Res. Labs., Kyoto, Japan

Over the past 40 years, x-ray films of the moving vocal tract have been the basis of much of the knowledge of speech production. Because of the risks involved with cineradiography, these films form an irreplaceable database. Unfortunately, as these films age, they deteriorate and become increasingly less accessible. In this presentation a collaborative effort between ATR Laboratories and Queen's University to transfer some of these films to videodisc is described. The goals were: (1) to archive the films in a stable medium that can serve as a database of high-quality vocal tract images suitable for image processing, (2) to make the films available by disseminating them to the speech community, and (3) to begin automatic extraction of time-varying descriptions of the full vocal tract. Currently, a 1 h videodisc has been produced. The disc contains 2--3 min, 35-mm high-speed films provided by J. Perkell (MIT) and C. Rochette (Universite Laval). The subjects were native speakers of Canadian English and Canadian French. The images were transferred to video frame by frame and the accompanying audio recordings have been synchronized with the images. The original audio recordings are also being made available separately on DAT.