Personalization Strategies for End-to-End Speech Recognition Systems
This addresses the challenge of personalized content recognition for speech recognition systems, offering incremental improvements over existing methods.
The paper tackles the problem of recognizing personalized content like contact names in end-to-end speech recognition by combining first-pass shallow fusion biasing and a novel second-pass de-biasing approach, achieving up to 16% improvement in personalized content recognition with minimal general degradation and an additional 14% improvement with second-pass de-biasing.
The recognition of personalized content, such as contact names, remains a challenging problem for end-to-end speech recognition systems. In this work, we demonstrate how first and second-pass rescoring strategies can be leveraged together to improve the recognition of such words. Following previous work, we use a shallow fusion approach to bias towards recognition of personalized content in the first-pass decoding. We show that such an approach can improve personalized content recognition by up to 16% with minimum degradation on the general use case. We describe a fast and scalable algorithm that enables our biasing models to remain at the word-level, while applying the biasing at the subword level. This has the advantage of not requiring the biasing models to be dependent on any subword symbol table. We also describe a novel second-pass de-biasing approach: used in conjunction with a first-pass shallow fusion that optimizes on oracle WER, we can achieve an additional 14% improvement on personalized content recognition, and even improve accuracy for the general use case by up to 2.5%.