Effect of Multi-band Frequency Compression for Improving Speech Perception in Monaural Hearing Aids on Source Localization
Jyoti M. Katagi, Pandurangarao N. Kulkarni · 2022
Normal hearing people use their auditory filters to separate sounds from various sound sources. But these auditory filters are widened in people with sensorineural hearing loss, so speech perception suffers as a result of this impairment, which also causes higher spectral masking. For decreasing the impact of spectrum masking, multiband frequency compression can be applied. In this method the frequency components in each of the analysis bands that make up the speech spectrum are compressed toward the center by a constant compression factor. For this procedure, we have used spectral segment mapping technique. Listening experiments used to evaluate these techniques revealed that spectral segment mapping provides desirable compression while preserving the spectral distribution of energy in the audio. In the present research work, we have calculated the impact of this method on source localization for a compression factor of 0.6, to improve speech perception for monaural hearing impaired listeners. The subjective analysis for 7 different azimuth angles (−90°, −60°, −30°, 0°, 30°, 60°, 90°) on 6 normal hearing listeners with different signal to noise ratio (SNR) conditions and 6 hearing impaired (moderate sensorineural hearing Loss (SNHL)) listeners shows that there is no adverse effect on source localization.