The effect of talker distribution on speech perception with beamformers in virtual audio-visual environments

Beamformers have the potential to greatly improve speech intelligibility for hearing aid users. However, most studies investigate beamformers in scenarios with source locations that are ideal for a beamformer and with static listeners. Here, we present a study that investigates the effect of beamformers in two reverberation conditions and with varying number of talkers, positioned either closely or widely spaced. Target locations were varied in the frontal hemisphere and participants could move their head freely. Audio-visual scenes were reproduced using a 64-channel loudspeaker array and virtual reality glasses. The listeners' task was to find a story in a mixture of other stories and correct identification and response time were used as outcome measures. The results show improved speech perception with beamformers in comparison to not using a beamformer. The beamformers' performance was found to be dependent on the distribution of the sources. Larger improvements of speech perception due to the beamformer were found when the reverberation time was higher. These findings demonstrate the potential of beamformers to improve speech perception in environments that are more like real-world scenes in comparison to previous studies.