@inproceedings{lin-etal-2025-neko,
    title = "{N}e{K}o: Cross-Modality Post-Recognition Error Correction with Tasks-Guided Mixture-of-Experts Language Model",
    author = "Lin, Yen-Ting  and
      Chen, Zhehuai  and
      Zelasko, Piotr  and
      Wan, Zhen  and
      Yang, Xuesong  and
      Chen, Zih-Ching  and
      Puvvada, Krishna C  and
      Hu, Ke  and
      Fu, Szu-Wei  and
      Chiu, Jun Wei  and
      Balam, Jagadeesh  and
      Ginsburg, Boris  and
      Wang, Yu-Chiang Frank  and
      Yang, Chao-Han Huck",
    editor = "Rehm, Georg  and
      Li, Yunyao",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/corrections-2025-08/2025.acl-industry.17/",
    doi = "10.18653/v1/2025.acl-industry.17",
    pages = "222--236",
    ISBN = "979-8-89176-288-6",
    abstract = "Construction of a general-purpose post-recognition error corrector poses a crucial question: how can we most effectively train a model on a large mixture of domain datasets? The answer would lie in learning dataset-specific features and digesting their knowledge in a single model. Previous methods achieve this by having separate correction language models, resulting in a significant increase in parameters. In this work, we present Mixture-of-Experts as a solution, highlighting that MoEs are much more than a scalability tool. We propose a Multi-Task Correction MoE, where we train the experts to become an ``expert'' of speech-to-text, language-to-text and vision-to-text datasets by learning to route each dataset{'}s tokens to its mapped expert. Experiments on the Open ASR Leaderboard show that we explore a new state-of-the-art performance by achieving an average relative 5.0{\%} WER reduction and substantial improvements in BLEU scores for speech and translation tasks. On zero-shot evaluation, NeKo outperforms GPT-3.5 and Claude-3.5-Sonnet with 15.5{\%} to 27.6{\%} relative WER reduction in the Hyporadise benchmark. NeKo performs competitively on grammar and post-OCR correction as a multi-task model."
}