Joint Audio/Text Training for Transformer Rescorer of Streaming Speech Recognition
This work addresses the data cost bottleneck for researchers and practitioners in speech recognition, though it is incremental as it builds on existing two-pass streaming ASR frameworks.
The paper tackles the problem of expensive paired audio-text data required for training Transformer Rescorer in streaming speech recognition by introducing a Joint Audio/Text training method that leverages cheaper unpaired text-only data, resulting in significant word error rate (WER) improvements on Librispeech and a large-scale in-house dataset without extra parameters or latency.
Recently, there has been an increasing interest in two-pass streaming end-to-end speech recognition (ASR) that incorporates a 2nd-pass rescoring model on top of the conventional 1st-pass streaming ASR model to improve recognition accuracy while keeping latency low. One of the latest 2nd-pass rescoring model, Transformer Rescorer, takes the n-best initial outputs and audio embeddings from the 1st-pass model, and then choose the best output by re-scoring the n-best initial outputs. However, training this Transformer Rescorer requires expensive paired audio-text training data because the model uses audio embeddings as input. In this work, we present our Joint Audio/Text training method for Transformer Rescorer, to leverage unpaired text-only data which is relatively cheaper than paired audio-text data. We evaluate Transformer Rescorer with our Joint Audio/Text training on Librispeech dataset as well as our large-scale in-house dataset and show that our training method can improve word error rate (WER) significantly compared to standard Transformer Rescorer without requiring any extra model parameters or latency.