Loudspeaker Beamforming to Enhance Speech Recognition Performance of Voice Driven Applications

TL;DR

This paper introduces a robust loudspeaker beamforming algorithm that enhances speech recognition in noisy environments by creating low-energy zones around devices, balancing recognition accuracy and audio quality.

Contribution

The paper presents a novel beamforming algorithm that uses a human auditory perception-based distortion measure to improve speech recognition performance in noisy scenarios.

Findings

Improved speech recognition accuracy in noisy environments.

Ability to reduce acoustic energy around devices at some quality cost.

Validated through simulations and real-world experiments.

Abstract

In this paper we propose a robust loudspeaker beamforming algorithm which is used to enhance the performance of voice driven applications in scenarios where the loudspeakers introduce the majority of the noise, e.g. when music is playing loudly. The loudspeaker beamformer modifies the loudspeaker playback signals to create a low-acoustic-energy region around the device that implements automatic speech recognition for a voice driven application (VDA). The algorithm utilises a distortion measure based on human auditory perception to limit the distortion perceived by human listeners. Simulations and real-world experiments show that the proposed loudspeaker beamformer improves the speech recognition performance in all tested scenarios. Moreover, the algorithm allows to further reduce the acoustic energy around the VDA device at the expense of reduced objective audio quality at the listener's location.