Predicting Compact Phrasal Rewrites with Large Language Models forAutomatic Speech Recognition Post Editing
Abstract: Large Language Models (LLMs) excel at rewriting tasks such as text style transfer
and grammatical error correction. Although the output in these tasks often significantly overlaps with the input, the decoding cost still increases with output length, regardless of the number of overlaps. By leveraging the overlap between the input and the output, Kaneko and Okazaki (2023) proposed model-agnostic edit span representations to compress the rewrites to save computation. They reported an output length reduction rate of nearly 80% with minimal accuracy impact in four rewriting tasks. In this paper, we propose alternative edit phrase representations inspired by phrase-based statistical machine translation. We systematically compare our phrasal representations with their span representation. We apply the LLM rewriting model to the task of Automatic Speech Recognition (ASR) post editing and show that our target-phrase-only edit representation has the best efficiency-accuracy trade-off. On the LibriSpeech test set,
our method closes 50-60% of the WER gap between the edit span model and the full rewrite model while losing only 10-20% of the length reduction rate of the edit span model.
Paper Type: Short
Research Area: Efficient/Low-Resource Methods for NLP
Research Area Keywords: Edit Operation, Large Language Model, Efficient Decoding
Contribution Types: NLP engineering experiment
Languages Studied: English
Submission Number: 1837
Loading