Server-side Rescoring of Spoken Entity-centric Knowledge Queries for Virtual Assistants
This work addresses the challenge of accurate knowledge integration for on-device virtual assistants, representing an incremental improvement in domain-specific ASR performance.
The paper tackles the problem of improving automatic speech recognition for entity-rich queries in virtual assistants by investigating server-side rescoring with various language models, resulting in significant word error rate improvements of 23%-35% on entity-centric query subpopulations.
On-device Virtual Assistants (VAs) powered by Automatic Speech Recognition (ASR) require effective knowledge integration for the challenging entity-rich query recognition. In this paper, we conduct an empirical study of modeling strategies for server-side rescoring of spoken information domain queries using various categories of Language Models (LMs) (N-gram word LMs, sub-word neural LMs). We investigate the combination of on-device and server-side signals, and demonstrate significant WER improvements of 23%-35% on various entity-centric query subpopulations by integrating various server-side LMs compared to performing ASR on-device only. We also perform a comparison between LMs trained on domain data and a GPT-3 variant offered by OpenAI as a baseline. Furthermore, we also show that model fusion of multiple server-side LMs trained from scratch most effectively combines complementary strengths of each model and integrates knowledge learned from domain-specific data to a VA ASR system.