# Select or Rewrite: Oracle-Guided Fine-Tuning for Generative ASR Post-Editing from N-Best Hypotheses

arXiv: https://arxiv.org/abs/2610.15056 · HTML: https://mlsanity.com/paper/2610.15056/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 25 references that are in the map.
- First posted on arXiv 10 Oct 2026.

## Builds on (most cited first, 20 of 20)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Common Voice: A Massively-Multilingual Speech Corpus](https://mlsanity.com/paper/1912.06670/) (2019)
- [On Using Monolingual Corpora in Neural Machine Translation](https://mlsanity.com/paper/1503.03535/) (2015)
- [Generative Speech Recognition Error Correction with Large Language Models and Task-Activating Prompting](https://mlsanity.com/paper/2309.15649/) (2023)
- [Can Generative Large Language Models Perform ASR Error Correction?](https://mlsanity.com/paper/2307.04172/) (2023)
- [N-best T5: Robust ASR Error Correction using Multiple Input Hypotheses and Constrained Decoding Space](https://mlsanity.com/paper/2303.00456/) (2023)
- [Large Language Models are Efficient Learners of Noise-Robust Speech Recognition](https://mlsanity.com/paper/2401.10446/) (2024)
- [ASR Error Correction using Large Language Models](https://mlsanity.com/paper/2409.09554/) (2024)
- [Listen Again and Choose the Right Answer: A New Paradigm for Automatic Speech Recognition with Large Language Models](https://mlsanity.com/paper/2405.10025/) (2024)
- [Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition](https://mlsanity.com/paper/2409.09785/) (2024)
- [Failing Forward: Improving Generative Error Correction for ASR with Synthetic Data and Retrieval Augmentation](https://mlsanity.com/paper/2410.13198/) (2024)
- [ProGRes: Prompted Generative Rescoring on ASR n-Best](https://mlsanity.com/paper/2409.00217/) (2024)
- [Contextual ASR Error Handling with LLMs Augmentation for Goal-Oriented Conversational AI](https://mlsanity.com/paper/2501.06129/) (2025)
- [Better Pseudo-labeling with Multi-ASR Fusion and Error Correction by SpeechLLM](https://mlsanity.com/paper/2506.11089/) (2025)
- [Evolutionary Prompt Design for LLM-Based Post-ASR Error Correction](https://mlsanity.com/paper/2407.16370/) (2024)
- [LLM-based Generative Error Correction for Rare Words with Synthetic Data and Phonetic Context](https://mlsanity.com/paper/2505.17410/) (2025)
- [Speech Recognition Rescoring with Large Speech-Text Foundation Models](https://mlsanity.com/paper/2409.16654/) (2024)
- [Listening and Seeing Again: Generative Error Correction for Audio-Visual Speech Recognition](https://mlsanity.com/paper/2501.04038/) (2025)
- [Chain of Correction for Full-text Speech Recognition with Large Language Models](https://mlsanity.com/paper/2504.01519/) (2025)

## Abstract (arXiv)

Large language models can post-edit ASR by using linguistic context to compare N-best hypotheses and rewrite the transcript when needed, often improving WER without retraining the recognizer. However, prompt-only correction is brittle: it may over-edit when top-1 is correct and hallucinate when the reference is absent, and often needs very large models. We propose oracle-guided fine-tuning that turns N-best lists into supervision via three curated pools: corrections (reference in the list but not top-1) to teach selection, regularizers (top-1 correct) to discourage unnecessary rewrites, and hard negatives (reference absent) to suppress hallucination. Fine-tuning on this data provides more effective supervision than random subset fine-tuning for learning a hybrid select-or-rewrite policy, with the clearest gains from around 3B parameters on the proprietary dataset but less conclusive results on CommonVoice, a divergence suggesting that public read-speech text may overlap with LLM pretraining and mask genuine post-editing gains.