Speaker adaptation for Wav2vec2 based dysarthric ASR
This work addresses speaker adaptation for dysarthric ASR, which is an incremental improvement in a domain-specific area.
The paper tackled the problem of dysarthric speech recognition by proposing a speaker adaptation network for fine-tuning wav2vec2 using fMLLR features, achieving a 57.72% word error rate for high severity cases on the UASpeech dataset.
Dysarthric speech recognition has posed major challenges due to lack of training data and heavy mismatch in speaker characteristics. Recent ASR systems have benefited from readily available pretrained models such as wav2vec2 to improve the recognition performance. Speaker adaptation using fMLLR and xvectors have provided major gains for dysarthric speech with very little adaptation data. However, integration of wav2vec2 with fMLLR features or xvectors during wav2vec2 finetuning is yet to be explored. In this work, we propose a simple adaptation network for fine-tuning wav2vec2 using fMLLR features. The adaptation network is also flexible to handle other speaker adaptive features such as xvectors. Experimental analysis show steady improvements using our proposed approach across all impairment severity levels and attains 57.72\% WER for high severity in UASpeech dataset. We also performed experiments on German dataset to substantiate the consistency of our proposed approach across diverse domains.