A Comprehensive Solution to Connect Speech Encoder and Large Language Model for ASR

TL;DR

This paper proposes a comprehensive approach to improve speech recognition by better connecting speech encoders with large language models, addressing fine-tuning, alignment, and insertion errors, with promising results on Librispeech.

Contribution

It introduces a novel matching loss for better modality alignment and explores efficient fine-tuning and inference methods to reduce errors in speech recognition systems.

Findings

Partially fine-tuning with LoRA is cost-effective.

Matching loss improves speech-text alignment.

Training methods reduce insertion errors.

Abstract

Recent works have shown promising results in connecting speech encoders to large language models (LLMs) for speech recognition. However, several limitations persist, including limited fine-tuning options, a lack of mechanisms to enforce speech-text alignment, and high insertion errors especially in domain mismatch conditions. This paper presents a comprehensive solution to address these issues. We begin by investigating more thoughtful fine-tuning schemes. Next, we propose a matching loss to enhance alignment between modalities. Finally, we explore training and inference methods to mitigate high insertion errors. Experimental results on the Librispeech corpus demonstrate that partially fine-tuning the encoder and LLM using parameter-efficient methods, such as LoRA, is the most cost-effective approach. Additionally, the matching loss improves modality alignment, enhancing performance. The proposed training and inference methods significantly reduce insertion errors.