Dual-mode ASR: Unify and Improve Streaming ASR with Full-context Modeling

TL;DR

This paper introduces Dual-mode ASR, a unified end-to-end speech recognition framework that enhances streaming ASR accuracy and latency by jointly training with full-context models, applicable to convolutional and transformer architectures.

Contribution

The work presents a novel unified training framework for streaming and full-context ASR, improving performance through weight sharing and knowledge distillation, and achieves state-of-the-art results.

Findings

Significant improvements in latency and accuracy for streaming ASR.

Effective application to both convolutional and transformer-based models.

Achieved new state-of-the-art results on LibriSpeech and MultiDomain datasets.

Abstract

Streaming automatic speech recognition (ASR) aims to emit each hypothesized word as quickly and accurately as possible, while full-context ASR waits for the completion of a full speech utterance before emitting completed hypotheses. In this work, we propose a unified framework, Dual-mode ASR, to train a single end-to-end ASR model with shared weights for both streaming and full-context speech recognition. We show that the latency and accuracy of streaming ASR significantly benefit from weight sharing and joint training of full-context ASR, especially with inplace knowledge distillation during the training. The Dual-mode ASR framework can be applied to recent state-of-the-art convolution-based and transformer-based ASR networks. We present extensive experiments with two state-of-the-art ASR networks, ContextNet and Conformer, on two datasets, a widely used public dataset LibriSpeech and a large-scale dataset MultiDomain. Experiments and ablation studies demonstrate that Dual-mode ASR not only simplifies the workflow of training and deploying streaming and full-context ASR models, but also significantly improves both emission latency and recognition accuracy of streaming ASR. With Dual-mode ASR, we achieve new state-of-the-art streaming ASR results on both LibriSpeech and MultiDomain in terms of accuracy and latency.