Electronically simulated loss of frequency resolution

Circuitry has been developed which degrades speech signals by reducing the spectral resolution available to the listener: each single frequency component in the original speech signal is replaced by a band of noise in the processed output. Measurements were made on the speech perception of normally hearing adult subjects listening to monosyllabic words via such a processing system. The effective resolution bandwidth available to the listener was increased by approximately 100 to 800 Hz, giving a drop of up to approximately 60% in whole-word scores. A three-channel design allowed different degrees of processing in the frequency regions 0 to 2 kHz, 2 to 4 kHz and 4 kHz upwards. Results show that speech frequencies in the range 0 to 2 kHz are more susceptible to this type of degradation than higher frequencies. Future development of this circuitry may allow simulation of the loss of frequency resolution associated with sensorineural hearing loss.