Arjan J. Bosman
Guido F. Smoorenburg
Lab. of Exptl. Audiol., Dept. of Otorhinolaryngol., Univ. Hospital Utrecht, Heidelberglaan 100, NL-3584 CX Utrecht, The Netherlands
For many profoundly hearing-impaired speech reading is the most important means of communication; amplified speech may provide, at best, additional information to speech reading. In an attempt to improve audio-visual communication, speech pattern elements that were thought to be most beneficial as a supplement to lip-reading were analyzed. The elements were fundamental frequency (f[sub 0]), voicing, and the first two formants (F[sub 1] and F[sub 2]). With these elements auditory stimuli were pro-duced with different encodings for voiced speech fragments. In the F[sub 1]F[sub 2] encoding both formants were represented by two adjacent harmonics of f[sub 0] around each formant frequency; in the F[sub 1] encoding by two harmonics of f[sub 0] around F[sub 1]. In the f[sub 0] encoding fundamental frequency was encoded as a single sine wave. In all cases white noise was used during voiceless speech. Out of 20 profoundly hearing-impaired subjects, all subjects but one showed a considerable improvement in lipreading scores when supplemented with an auditory stimulus. In most cases highest scores were obtained with the natural speech condition. For four subjects the F[sub 1]-encoded signal provided highest scores. Relations among intelligibility scores and dynamic range, gap detection, and difference limen for frequency will be discussed.