Self-Training for End-to-End Speech Recognition
This work addresses speech recognition accuracy for noisy and clean speech settings, presenting a strong incremental improvement over previous methods.
The paper tackles improving end-to-end speech recognition accuracy by revisiting self-training with pseudo-labels, resulting in a 33.9% relative WER improvement in noisy speech and recovering 59.3% of the gap to an oracle model in clean speech.
We revisit self-training in the context of end-to-end speech recognition. We demonstrate that training with pseudo-labels can substantially improve the accuracy of a baseline model. Key to our approach are a strong baseline acoustic and language model used to generate the pseudo-labels, filtering mechanisms tailored to common errors from sequence-to-sequence models, and a novel ensemble approach to increase pseudo-label diversity. Experiments on the LibriSpeech corpus show that with an ensemble of four models and label filtering, self-training yields a 33.9% relative improvement in WER compared with a baseline trained on 100 hours of labelled data in the noisy speech setting. In the clean speech setting, self-training recovers 59.3% of the gap between the baseline and an oracle model, which is at least 93.8% relatively higher than what previous approaches can achieve.