Joint Speaker Features Learning for Audio-visual Multichannel Speech Separation and Recognition

TL;DR

This paper introduces joint speaker feature learning methods that enhance audio-visual multichannel speech separation and recognition, achieving significant WER reductions through integrated speaker encoders and fusion techniques.

Contribution

It presents a novel joint speaker feature learning approach that improves speech separation and recognition performance in zero-shot settings by integrating xVector and ECAPA-TDNN encoders.

Findings

Consistent performance improvements over baselines.

Significant WER reductions of over 20%.

Enhanced inter-speaker discrimination correlates with better results.

Abstract

This paper proposes joint speaker feature learning methods for zero-shot adaptation of audio-visual multichannel speech separation and recognition systems. xVector and ECAPA-TDNN speaker encoders are connected using purpose-built fusion blocks and tightly integrated with the complete system training. Experiments conducted on LRS3-TED data simulated multichannel overlapped speech suggest that joint speaker feature learning consistently improves speech separation and recognition performance over the baselines without joint speaker feature estimation. Further analyses reveal performance improvements are strongly correlated with increased inter-speaker discrimination measured using cosine similarity. The best-performing joint speaker feature learning adapted system outperformed the baseline fine-tuned WavLM model by statistically significant WER reductions of 21.6% and 25.3% absolute (67.5% and 83.5% relative) on Dev and Test sets after incorporating WavLM features and video modality.