Noise-Robust AV-ASR Using Visual Features Both in the Whisper Encoder and Decoder

TL;DR

This paper introduces a dual-use visual feature fusion method in Whisper AV-ASR models, significantly improving noise robustness and establishing new state-of-the-art results in noisy conditions.

Contribution

Proposes a novel dual-use visual fusion approach in Whisper models, enhancing noise robustness and outperforming existing fusion methods in AV-ASR.

Findings

35% relative WER reduction in Whisper small model

57% relative WER reduction in Whisper medium model

Achieves state-of-the-art results on LRS3 benchmark

Abstract

In audiovisual automatic speech recognition (AV-ASR) systems, information fusion of visual features in a pre-trained ASR has been proven as a promising method to improve noise robustness. In this work, based on the prominent Whisper ASR, first, we propose a simple and effective visual fusion method -- use of visual features both in encoder and decoder (dual-use) -- to learn the audiovisual interactions in the encoder and to weigh modalities in the decoder. Second, we compare visual fusion methods in Whisper models of various sizes. Our proposed dual-use method shows consistent noise robustness improvement, e.g., a 35% relative improvement (WER: 4.41% vs. 6.83%) based on Whisper small, and a 57% relative improvement (WER: 4.07% vs. 9.53%) based on Whisper medium, compared to typical reference middle fusion in babble noise with a signal-to-noise ratio (SNR) of 0dB. Third, we conduct ablation studies examining the impact of various module designs and fusion options. Fine-tuned on 1929 hours of audiovisual data, our dual-use method using Whisper medium achieves 4.08% (MUSAN babble noise) and 4.43% (NoiseX babble noise) average WER across various SNRs, thereby establishing a new state-of-the-art in noisy conditions on the LRS3 AV-ASR benchmark. Our code is at https://github.com/ifnspaml/Dual-Use-AVASR