Developing RNN-T Models Surpassing High-Performance Hybrid Models with Customization Capability
This work addresses the problem of improving and deploying end-to-end speech recognition models for practical applications, representing an incremental advancement with specific gains.
The authors developed an RNN-T model that surpasses a high-performance hybrid model in speech recognition accuracy and latency when trained on 65,000 hours of data, and they identified text-to-speech generation as the most effective method for customizing the model to new domains.
Because of its streaming nature, recurrent neural network transducer (RNN-T) is a very promising end-to-end (E2E) model that may replace the popular hybrid model for automatic speech recognition. In this paper, we describe our recent development of RNN-T models with reduced GPU memory consumption during training, better initialization strategy, and advanced encoder modeling with future lookahead. When trained with Microsoft's 65 thousand hours of anonymized training data, the developed RNN-T model surpasses a very well trained hybrid model with both better recognition accuracy and lower latency. We further study how to customize RNN-T models to a new domain, which is important for deploying E2E models to practical scenarios. By comparing several methods leveraging text-only data in the new domain, we found that updating RNN-T's prediction and joint networks using text-to-speech generated from domain-specific text is the most effective.