Peeking Into The Future For Contextual Biasing
This addresses a critical issue for virtual assistants and similar applications by improving recognition of rare entities, though it is an incremental advance in contextual biasing methods.
The paper tackles the problem of end-to-end ASR models struggling with rare named entities by proposing a contextual biasing method that predicts multiple future tokens to score candidates, achieving up to 50.34% relative improvement in named entity word error rate on Librispeech.
While end-to-end (E2E) automatic speech recognition (ASR) models excel at general transcription, they struggle to recognize rare or unseen named entities (e.g., contact names, locations), which are critical for downstream applications like virtual assistants. In this paper, we propose a contextual biasing method for attention based encoder decoder (AED) models using a list of candidate named entities. Instead of predicting only the next token, we simultaneously predict multiple future tokens, enabling the model to "peek into the future" and score potential candidate entities in the entity list. Moreover, our approach leverages the multi-token prediction logits directly without requiring additional entity encoders or cross-attention layers, significantly reducing architectural complexity. Experiments on Librispeech demonstrate that our approach achieves up to 50.34% relative improvement in named entity word error rate compared to the baseline AED model.