Effect of Binaural Dichotic Presentation with Critical Bandwidth Based Comb Filters on Source Localization

Earlier studies have shown that binaural dichotic presentation, using critical bandwidth based spectral splitting with perceptually balanced comb filters, helps in reducing the effect of spectral masking for persons with moderate bilateral sensorineural hearing impairment. It has been reported to provide an SNR advantage of 5 dB for normal hearing subjects under simulated hearing loss, and a relative improvement of 6-121 % in recognition scores for hearing impaired listeners. The objective of the present study is to investigate the effect of this type of binaural dichotic presentation on source localization. Investigations were carried out by processing the stimuli with head related transfer functions (HRTFs) for the two ears with binaural presentation using headphones, for 0 degree elevation and azimuth angles varying from -90 to +90 degrees. The test material included sound of breaking glass, VCV syllable /aba/, and noise band-pass filtered with different bandwidths. A considerable degradation in source localization was observed for narrow band stimuli. But the subjects were able to perceive the source direction for stimuli with bandwidth exceeding one-third octave, indicating that use of dichotic presentation should not impair source localization for speech and broadband environmental sounds. INTRODUCTION In cases of sensorineural hearing loss, the auditory filter bandwidth generally increases and frequency selectivity gets reduced due to increased masking [1] [3]. For reducing the effect of increased spectral masking for listeners with moderate bilateral loss, several studies have examined the use of binaural dichotic presentation, with speech signal split into two parts with complimentary spectra with adjacent bands presented to different ears [4] [9]. It has been reported that use of auditory critical bandwidth based spectral splitting technique using perceptually balanced comb filters helps in reducing the effect of spectral masking for persons with moderate bilateral sensorineural hearing loss. Using this technique, Cheeran and Pandey [7], [8] reported an SNR advantage of 5 dB for normal hearing subjects under simulated hearing loss, and a relative improvement of 6-121 % in recognition scores and 6.4-312.5 % in relative information transmission of consonantal place feature for hearing impaired listeners. As this signal processing presents the information dichotically, a study of its effect on source localization is very important. Accurate localization of sound source is a complex perceptual process and it involves integration of multiple acoustic cues [1]. The most important cues for source localization in the horizontal plane are inter-aural time difference (ITD) and inter-aural level difference (ILD) [10], [11]. According to the “duplex” theory of binaural localization, ITD is important for localization at low frequencies, while high frequencies are localized from ILD [12]. Localization in vertical plane and discrimination of back and front primarily depends upon high frequency (> 5 kHz) spectral cues that are created by reflection and diffraction of sound by the external ear (particularly pinna) [13]. ITD can be calculated from the path length difference between the two ears and it varies from a minimum value of 0 for a sound coming from straight ahead, to a maximum value of 690 μs for a sound coming from a source located directly opposite to one ear [13]. ILD varies over 0-20 dB [14]. Best et al. [11] in their study on the role of high frequencies in source localization showed that preserving the information above 8 kHz helps in accurate localization of elevation. In spectral splitting for dichotic presentation, each band gets presented monaurally and hence source localization ability may get highly impaired for narrow band signals. For