When people who wear hearing aids speak, there are three paths by which they hear their own voices: 1) through the air and leakage around the earmold; 2) via the solid structures of their head; 3) through the air to the hearing aid microphone, and then through the aid circuitry. These paths involve different time delays. Digital processing introduces delays in path 3 from a few to several tens of milliseconds, which could lead to a range of disturbing effects. We examined one purely auditory effect, namely hearing speech through all three of these paths. Subjective disturbance was measured as a function of delay in path 3 using simulations of hearing loss and a simulated hearing aid. With increasing hearing loss, the loudness of sound heard via paths 1 and 2 decreases, and the aid user relies more on path 3. The disturbance produced by the delay then might be less perceptible. To test this idea, four different hearing losses were simulated, varying from mild to moderately severe.
Each of two talkers was fitted with a closed earmold, and simultaneous above-ear and in-ear recordings were made of each talker reading prose. The above-ear signal was amplified using a simulated hearing aid with 4-channel full dynamic range compression; compression ratios and gains were selected using an algorithm based on the absolute thresholds used in the simulations of hearing loss. The resultant output was then mixed with the in-ear signal with one of five values of delay, and the combined signal was processed using the four simulations of hearing loss. The resulting stimuli simulated for normal-hearing listeners the experience of having a hearing impairment and listening through a hearing aid while talking, except that the talker’s voice was not that of the listener. Twenty normally hearing subjects gave subjective ratings of the disturbance of the echo for each delay and each simulated hearing loss.
Disturbance ratings generally increased monotonically with increasing delay. Average results show that delays are rated as “disturbing” for values between 20 and 30 msec for mild to moderate losses. For a moderately severe loss, the rating “disturbing” was not quite achieved at 40 msec. For moderate losses, a speaker with low fundamental frequency (f0
)(70 to 140 Hz) was less disturbing than a speaker with a medium f0
, (100 to 180 Hz). This effect reversed for the mildest loss for low values of delay.
The auditory effects of delays between bone-conducted sound and aid-conducted sound are likely to become disturbing for delays exceeding 20 msec. Somewhat longer delays may be tolerable for moderate to severe hearing losses. These delays are smaller than the delays at which audio-visual integration is disrupted.