Generative Speech Recognition Error Correction with Large Language Models and Task-Activating Prompting
This addresses speech recognition accuracy for applications requiring robust performance across domains, though it is incremental in applying LLMs to this specific task.
The authors tackled speech recognition error correction by using large language models (LLMs) as post-processors with prompting techniques, achieving competitive results with domain-tuned models on out-of-domain tasks and error rates below N-best oracle levels when combined with fine-tuning.
We explore the ability of large language models (LLMs) to act as speech recognition post-processors that perform rescoring and error correction. Our first focus is on instruction prompting to let LLMs perform these task without fine-tuning, for which we evaluate different prompting schemes, both zero- and few-shot in-context learning, and a novel task activation prompting method that combines causal instructions and demonstration to increase its context windows. Next, we show that rescoring only by in-context learning with frozen LLMs achieves results that are competitive with rescoring by domain-tuned LMs, using a pretrained first-pass recognition system and rescoring output on two out-of-domain tasks (ATIS and WSJ). By combining prompting techniques with fine-tuning we achieve error rates below the N-best oracle level, showcasing the generalization power of the LLMs.