Adaptive Contextual Biasing for Transducer Based Streaming Speech Recognition

TL;DR

This paper introduces an adaptive biasing method for streaming speech recognition that dynamically switches contextual biasing on and off, improving accuracy for both personalized and common words with minimal latency impact.

Contribution

The proposed method uses a Context-Aware Transformer Transducer to adaptively control biasing, significantly reducing errors in both personalized and common word recognition scenarios.

Findings

Achieves up to 6.7% and 20.7% relative WER and CER reduction on datasets.

Mitigates up to 96.7% and 84.9% of error increase for common cases.

Maintains minimal RTF increase with negligible performance impact.

Abstract

By incorporating additional contextual information, deep biasing methods have emerged as a promising solution for speech recognition of personalized words. However, for real-world voice assistants, always biasing on such personalized words with high prediction scores can significantly degrade the performance of recognizing common words. To address this issue, we propose an adaptive contextual biasing method based on Context-Aware Transformer Transducer (CATT) that utilizes the biased encoder and predictor embeddings to perform streaming prediction of contextual phrase occurrences. Such prediction is then used to dynamically switch the bias list on and off, enabling the model to adapt to both personalized and common scenarios. Experiments on Librispeech and internal voice assistant datasets show that our approach can achieve up to 6.7% and 20.7% relative reduction in WER and CER compared to the baseline respectively, mitigating up to 96.7% and 84.9% of the relative WER and CER increase for common cases. Furthermore, our approach has a minimal performance impact in personalized scenarios while maintaining a streaming inference pipeline with negligible RTF increase.