A tactually cued vocoder using a microprocessor

In order to improve the recognition of consonants by our tactual vocoder [T. Ifukube, 9th ICA, I‐72 (1977)], a tactually cued vocoder was designed for the deaf. In this apparatus, the acoustic cues of Japanese monosyllables were extracted automatically in real time using a microprocessor. These cues were displayed on a small vibrator matrix with 16 × 6 piezoelectric devices so as to be perceivable by the vibrotactile sense of a fingertip. Experimental results showed that unvoiced consonants, except plosives, could be detected at about 90% correct rate. Sixty‐eight Japanese monosyllables were separated into three groups (fricatives, plosives, and others) and were detected at a rate of about 80% correct. The possibilities for the practical use of our cued vocoder were discussed based on the results of tactile speech recognition by the deaf.