988368

Objectives: To evaluate different signal-processing algorithms for tactile identification of environmental sounds in a monitoring aid for the deafblind. Two men and three women, sensorineurally deaf or profoundly hearing impaired with experience of vibratory experiments, age 22-36 years.

Methods: A closed set of 45 representative environmental sounds were processed using two transposing (TRHA, TR1/3) and three modulating algorithms (AM, AMFM, AMMC) and presented as tactile stimuli using a portable vibrator in three experiments. The algorithms TRHA, TR1/3, AMFM and AMMC had two alternatives (with and without adaption to vibratory thresholds). In Exp. 1, the sounds were preprocessed and directly fed to the vibrator. In Exp. 2 and 3, the sounds were presented in an acoustic test room, without or with background noise (SNR=+5 dB), and processed in real time.

Results: In Exp. 1, Algorithm AMFM and AMFM(A) consistently had the lowest identification scores, and were thus excluded in Exp. 2 and 3. TRHA, AM, AMMC, and AMMC(A) showed comparable identification scores (30%-42%) and the addition of noise did not deteriorate the performance.

Discussion: Algorithm TRHA, AM, AMMC, and AMMC(A) showed good performance in all three experiments and were robust in noise they can therefore be used in further testing in real environments.