Reference Microphone Selection for Guided Source Separation based on the Normalized L-p Norm

TL;DR

This paper introduces novel reference microphone selection methods for Guided Source Separation that leverage the normalized L-p norm, improving speech enhancement and ASR performance in distributed microphone setups.

Contribution

It proposes two new microphone selection techniques based on the normalized L-p norm, combining SNR and ELR considerations for better speech enhancement.

Findings

L-p norm-based methods outperform baseline in reducing word error rate

Proposed methods effectively balance SNR and reverberation differences

Improved ASR performance demonstrated on CHiME-8 dataset

Abstract

Guided Source Separation (GSS) is a popular front-end for distant automatic speech recognition (ASR) systems using spatially distributed microphones. When considering spatially distributed microphones, the choice of reference microphone may have a large influence on the quality of the output signal and the downstream ASR performance. In GSS-based speech enhancement, reference microphone selection is typically performed using the signal-to-noise ratio (SNR), which is optimal for noise reduction but may neglect differences in early-to-late-reverberant ratio (ELR) across microphones. In this paper, we propose two reference microphone selection methods for GSS-based speech enhancement that are based on the normalized $\ell_p$-norm, either using only the normalized $\ell_p$-norm or combining the normalized $\ell_p$-norm and the SNR to account for both differences in SNR and ELR across microphones. Experimental evaluation using a CHiME-8 distant ASR system shows that the proposed $\ell_p$-norm-based methods outperform the baseline method, reducing the macro-average word error rate.