Adapting large language model (LLM)-based automatic speech recognition (ASR) systems to new domains using text-only data is a significant yet underexplored challenge. Standard fine-tuning of the LLM on the target domain text often disrupts the critical alignment between the speech and text modality learned by the projector, degrading performance. We introduce a novel text-only adaptation method that frames this process as a text denoising task. Our approach trains the LLM to recover clean transcripts from noisy inputs. This process effectively adapts the model to a target domain while preserving cross-modal alignment. Our solution is lightweight, requiring no architectural changes or additional parameters. Extensive evaluation on two datasets demonstrates up to 22.1% relative improvement, outperforming recent state-of-the-art text-only adaptation methods.
@article{arxiv.2601.20900,
title = {Text-only adaptation in LLM-based ASR through text denoising},
author = {Andrés Carofilis and Sergio Burdisso and Esaú Villatoro-Tello and Shashi Kumar and Kadri Hacioglu and Srikanth Madikeri and Pradeep Rangappa and Manjunath K E and Petr Motlicek and Shankar Venkatesan and Andreas Stolcke},
journal= {arXiv preprint arXiv:2601.20900},
year = {2026}
}