IANS: Intelligibility-aware Null-steering Beamforming for Dual-Microphone Arrays

TL;DR

This paper introduces IANS, a novel speech enhancement framework that uses intelligibility prediction to optimize null-steering beamforming for dual-microphone arrays without prior speech parameter knowledge.

Contribution

The paper proposes a new intelligibility-aware null-steering beamforming method that leverages STOI-Net for improved speech clarity without needing prior DOA or RTF information.

Findings

IANS improves speech intelligibility in dual-microphone setups.

Performance is comparable to traditional methods with known DOA.

The approach enhances speech clarity without prior speech parameter knowledge.

Abstract

Beamforming techniques are popular in speech-related applications due to their effective spatial filtering capabilities. Nonetheless, conventional beamforming techniques generally depend heavily on either the target's direction-of-arrival (DOA), relative transfer function (RTF) or covariance matrix. This paper presents a new approach, the intelligibility-aware null-steering (IANS) beamforming framework, which uses the STOI-Net intelligibility prediction model to improve speech intelligibility without prior knowledge of the speech signal parameters mentioned earlier. The IANS framework combines a null-steering beamformer (NSBF) to generate a set of beamformed outputs, and STOI-Net, to determine the optimal result. Experimental results indicate that IANS can produce intelligibility-enhanced signals using a small dual-microphone array. The results are comparable to those obtained by null-steering beamformers with given knowledge of DOAs.