Contextualization of ASR with LLM using phonetic retrieval-based augmentation

TL;DR

This paper introduces a retrieval-based method to improve speech recognition by contextualizing large language models with phonetic retrieval of personal named entities, significantly reducing errors in voice assistant tasks.

Contribution

It proposes a novel retrieval-based approach that enhances LLM-based speech recognition by incorporating phonetic retrieval of personal entities, improving accuracy without large prompting overhead.

Findings

Up to 30.2% relative word error rate reduction.

73.6% relative named entity error rate reduction.

Efficient handling of large personal databases.

Abstract

Large language models (LLMs) have shown superb capability of modeling multimodal signals including audio and text, allowing the model to generate spoken or textual response given a speech input. However, it remains a challenge for the model to recognize personal named entities, such as contacts in a phone book, when the input modality is speech. In this work, we start with a speech recognition task and propose a retrieval-based solution to contextualize the LLM: we first let the LLM detect named entities in speech without any context, then use this named entity as a query to retrieve phonetically similar named entities from a personal database and feed them to the LLM, and finally run context-aware LLM decoding. In a voice assistant task, our solution achieved up to 30.2% relative word error rate reduction and 73.6% relative named entity error rate reduction compared to a baseline system without contextualization. Notably, our solution by design avoids prompting the LLM with the full named entity database, making it highly efficient and applicable to large named entity databases.