The effect of an interaural time difference to level difference transformation method on speech intelligibility in a multi-talker setting.

Abstract

The human auditory system is able to improve speech intelligibility in challenging multi-talker scenarios by using the binaural cues [Interaural Time Differences (ITDs) and Interaural Level Differences (ILDs)] for spatially separated talkers. However, aging and hearing loss can vastly degrade a listener's ability to perceive ITDs, resulting in a decreased binaural benefit. For this reason, we investigated a method to transform imperceptible ITDs (artificially removed to simulate ITD sensitivity loss) into low-frequency ILDs to assess the binaural benefit that can be gained this way, based on listening experiments with normal-hearing listeners. In a reverberant environment, no binaural benefit was found. In an anechoic environment and with a central target speaker, the decrease in speech reception tresholds because of removed ITDs was almost completely compensated when low-frequency ILDs were added. For a lateral target speaker, the low-frequency ILDs even provided a significant improvement over the baseline where ITDs and ILDs are available. Simulations using the binaural speech intelligibility model (BSIM) supported the measured results from the listening experiment. These findings suggest that this cue conversion method could potentially bring a benefit to hearing impaired and elderly listeners when implemented in cochlear implants and hearing aids.