ASCLSDSep 25, 2023

Connecting Speech Encoder and Large Language Model for ASR

arXiv:2309.13963v2124 citationsh-index: 21
Originality Incremental advance
AI Analysis

This work addresses the challenge of building efficient ASR systems by connecting speech and language models, offering incremental improvements in accuracy and generalization for speech recognition tasks.

The paper tackled the problem of integrating speech encoders with large language models for automatic speech recognition by comparing three connector structures, finding that Q-Former-based models achieved consistent word error rate reductions, such as 12% relative improvement on out-of-domain data and 17% on long speech segments.

The impressive capability and versatility of large language models (LLMs) have aroused increasing attention in automatic speech recognition (ASR), with several pioneering studies attempting to build integrated ASR models by connecting a speech encoder with an LLM. This paper presents a comparative study of three commonly used structures as connectors, including fully connected layers, multi-head cross-attention, and Q-Former. Speech encoders from the Whisper model series as well as LLMs from the Vicuna model series with different model sizes were studied. Experiments were performed on the commonly used LibriSpeech, Common Voice, and GigaSpeech datasets, where the LLMs with Q-Formers demonstrated consistent and considerable word error rate (WER) reductions over LLMs with other connector structures. Q-Former-based LLMs can generalise well to out-of-domain datasets, where 12% relative WER reductions over the Whisper baseline ASR model were achieved on the Eval2000 test set without using any in-domain training data from Switchboard. Moreover, a novel segment-level Q-Former is proposed to enable LLMs to recognise speech segments with a duration exceeding the limitation of the encoders, which results in 17% relative WER reductions over other connector structures on 90-second-long speech data.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes