Improving Neural Biasing for Contextual Speech Recognition by Early Context Injection and Text Perturbation
This work addresses the challenge of recognizing rare words and named entities in speech recognition for users and applications, representing an incremental improvement over existing biasing methods.
The paper tackled the problem of improving context-aware automatic speech recognition by proposing early context injection and text perturbation techniques, resulting in a 60% relative reduction in rare word error rate on LibriSpeech compared to no biasing and achieving state-of-the-art performance.
Existing research suggests that automatic speech recognition (ASR) models can benefit from additional contexts (e.g., contact lists, user specified vocabulary). Rare words and named entities can be better recognized with contexts. In this work, we propose two simple yet effective techniques to improve context-aware ASR models. First, we inject contexts into the encoders at an early stage instead of merely at their last layers. Second, to enforce the model to leverage the contexts during training, we perturb the reference transcription with alternative spellings so that the model learns to rely on the contexts to make correct predictions. On LibriSpeech, our techniques together reduce the rare word error rate by 60% and 25% relatively compared to no biasing and shallow fusion, making the new state-of-the-art performance. On SPGISpeech and a real-world dataset ConEC, our techniques also yield good improvements over the baselines.