Encoding voice fundamental frequency into vibrotactile frequency

Abstract

The ability of 8 hearing and 5 deaf human subjects was measured to identify the stress pattern in a short sentence from the variation in voice fundamental frequency (F0), when presented aurally (for hearing subjects) and when transformed into vibrotactile pulse frequency. Various transformations from F0 to pulse frequency were tested in an attempt to determine an optimum transformation, the amount of F0 information that could be transmitted and what the limitations in the tactile channel might be. A 1- or 2-octave reduction of F0 vibrotactile frequency (transmitting every 2nd or 3rd glottal pulse) might result in a significant ability to discriminate the intonation patterns associated with moderate-to-strong patterns of sentence stress in English. Accurate reception of the details of the intonation pattern may require a slower than normal pronunciation because of an apparent temporal indeterminacy of about 200 ms in the perception of variations in vibrotactile frequency. A performance deficit noted for the 2 prelingually, profoundly deaf subjects with marginally discriminable encodings offers some support for the previous hypothesis that there is a natural association between auditory pitch and perceived vibrotactile frequency.