Lipreading with vibrotactile vocoders

A training study was conducted to compare aided lipreading of normal‐hearing and deaf adults, each assigned to one of three vibrotactile vocoders. Vocoders were (1) the Queen's University/Central Institute for the Deaf (QU/CID) vocoder, with one‐third octave filter spacing and logarithmic output scaling; (2) the QU/CID vocoder with linear output scaling; and (3) the GU (Gallaudet University) vocoder designed for greater resolution than the others in the F2 region and linear output scaling. Subjects received stimuli in baseline (no vocoder) and treatment (vocoder) conditions. In addition, two subjects served as visual‐only controls. Stimuli were provided by a live talker and two talkers prerecorded on laser videodisc (Bernstein and Eberhardt, 1986). Preliminary analysis of the results suggests that (1) the QU/CID linear vocoder was most effective, followed by the GU vocoder with linear output; and (2) regardless of experimental condition, normal‐hearing subjects' lipreading improved over the approximately ...