Improving Tail Performance of a Deliberation E2E ASR Model Using a Large Text Corpus
This work addresses the challenge of recognizing infrequent words in ASR systems, which is important for improving accuracy in real-world applications, though it is incremental as it builds on existing shallow fusion methods.
The authors tackled the problem of improving tail word recognition in end-to-end ASR models by incorporating a large text corpus via shallow fusion, showing that pruning the training set is more effective than increasing model size and that MWER fine-tuning reduces hyperparameter sensitivity.
End-to-end (E2E) automatic speech recognition (ASR) systems lack the distinct language model (LM) component that characterizes traditional speech systems. While this simplifies the model architecture, it complicates the task of incorporating text-only data into training, which is important to the recognition of tail words that do not occur often in audio-text pairs. While shallow fusion has been proposed as a method for incorporating a pre-trained LM into an E2E model at inference time, it has not yet been explored for very large text corpora, and it has been shown to be very sensitive to hyperparameter settings in the beam search. In this work, we apply shallow fusion to incorporate a very large text corpus into a state-of-the-art E2EASR model. We explore the impact of model size and show that intelligent pruning of the training set can be more effective than increasing the parameter count. Additionally, we show that incorporating the LM in minimum word error rate (MWER) fine tuning makes shallow fusion far less dependent on optimal hyperparameter settings, reducing the difficulty of that tuning problem.