Interactive Feature Fusion for End-to-End Noise-Robust Speech Recognition
This addresses noise-robust speech recognition for ASR systems, but it is incremental as it builds on existing feature fusion methods.
The paper tackled the problem of over-suppression in speech enhancement degrading ASR performance by proposing an interactive feature fusion network (IFF-Net) to learn complementary information from enhanced and noisy features, achieving a 4.1% absolute WER reduction on the RATS Channel-A corpus.
Speech enhancement (SE) aims to suppress the additive noise from a noisy speech signal to improve the speech's perceptual quality and intelligibility. However, the over-suppression phenomenon in the enhanced speech might degrade the performance of downstream automatic speech recognition (ASR) task due to the missing latent information. To alleviate such problem, we propose an interactive feature fusion network (IFF-Net) for noise-robust speech recognition to learn complementary information from the enhanced feature and original noisy feature. Experimental results show that the proposed method achieves absolute word error rate (WER) reduction of 4.1% over the best baseline on RATS Channel-A corpus. Our further analysis indicates that the proposed IFF-Net can complement some missing information in the over-suppressed enhanced feature.