# Listen to the Latents: Self-Correcting Speech Recognition in Large Audio Language Models Through Hidden-State Interactions

arXiv: https://arxiv.org/abs/2609.02940 · HTML: https://mlsanity.com/paper/2609.02940/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 44 references that are in the map.
- First posted on arXiv 31 Aug 2026.

## Builds on (most cited first, 28 of 28)

- [Bootstrap your own latent: A new approach to self-supervised Learning](https://mlsanity.com/paper/2006.07733/) (2020)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Deep contextualized word representations](https://mlsanity.com/paper/1802.05365/) (2018)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://mlsanity.com/paper/1909.08053/) (2019)
- [Common Voice: A Massively-Multilingual Speech Corpus](https://mlsanity.com/paper/1912.06670/) (2019)
- [VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation Learning, Semi-Supervised Learning and Interpretation](https://mlsanity.com/paper/2101.00390/) (2021)
- [GigaSpeech: An Evolving, Multi-domain ASR Corpus with 10,000 Hours of Transcribed Audio](https://mlsanity.com/paper/2106.06909/) (2021)
- [TED-LIUM 3: twice as much data and corpus repartition for experiments on speaker adaptation](https://mlsanity.com/paper/1805.04699/) (2018)
- [SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition](https://mlsanity.com/paper/2104.02014/) (2021)
- [SALM: Speech-augmented Language Model with In-context Learning for Speech Recognition and Translation](https://mlsanity.com/paper/2310.09424/) (2023)
- [Generative Speech Recognition Error Correction with Large Language Models and Task-Activating Prompting](https://mlsanity.com/paper/2309.15649/) (2023)
- [YODAS: Youtube-Oriented Dataset for Audio and Speech](https://mlsanity.com/paper/2406.00899/) (2024)
- [Latent Space Chain-of-Embedding Enables Output-free LLM Self-Evaluation](https://mlsanity.com/paper/2410.13640/) (2024)
- [Whispering LLaMA: A Cross-Modal Generative Error Correction Framework for Speech Recognition](https://mlsanity.com/paper/2310.06434/) (2023)
- [Earnings-22: A Practical Benchmark for Accents in the Wild](https://mlsanity.com/paper/2203.15591/) (2022)
- [N-best T5: Robust ASR Error Correction using Multiple Input Hypotheses and Constrained Decoding Space](https://mlsanity.com/paper/2303.00456/) (2023)
- [OWSM v4: Improving Open Whisper-Style Speech Models via Data Scaling and Cleaning](https://mlsanity.com/paper/2506.00338/) (2025)
- [Reducing Object Hallucination in Large Audio-Language Models via Audio-Aware Decoding](https://mlsanity.com/paper/2506.07233/) (2025)
- [Open ASR Leaderboard: Towards Reproducible and Transparent Multilingual and Long-Form Speech Recognition Evaluation](https://mlsanity.com/paper/2510.06961/) (2025)
- [Causal Estimation of Tokenisation Bias](https://mlsanity.com/paper/2506.03149/) (2025)
- [Domain-aware Neural Language Models for Speech Recognition](https://mlsanity.com/paper/2101.03229/) (2021)
- [Zero-shot Domain-sensitive Speech Recognition with Prompt-conditioning Fine-tuning](https://mlsanity.com/paper/2307.10274/) (2023)
- [SALSA: Speedy ASR-LLM Synchronous Aggregation](https://mlsanity.com/paper/2408.16542/) (2024)
- [NeKo: Cross-Modality Post-Recognition Error Correction with Tasks-Guided Mixture-of-Experts Language Model](https://mlsanity.com/paper/2411.05945/) (2024)
- [A Self-Refining Framework for Enhancing ASR Using TTS-Synthesized Data](https://mlsanity.com/paper/2506.11130/) (2025)
- [Let's Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR](https://mlsanity.com/paper/2405.14259/) (2024)
- [Spoken Conversational Agents with Large Language Models](https://mlsanity.com/paper/2512.02593/) (2025)

## Abstract (arXiv)

Recent automatic speech recognition (ASR) systems increasingly integrate large language models (LLMs) to leverage their semantic knowledge, either externally through logit fusion or internally through warm initialization. However, how to effectively combine these two strategies remains underexplored. In this work, we refine warm-initialized LLM-based ASR models by leveraging their own pre-adaptation base LLMs, focusing on LoRA-adapted settings where the base LLM is preserved. To achieve this, we propose Hybrid Search, a targeted correction strategy motivated by two observations. First, interaction features that characterize the relationship between LLM-based ASR hidden states and base-LLM hidden states provide informative signals about a token's degree of semantic dependence. Second, selectively refining targeted tokens with high semantic dependence improves ASR performance far beyond naive global LLM-correction methods including rescoring and late fusion. Our analysis suggests that, even after semantic knowledge transfer through warm initialization, LLM-based ASR models can still leverage their base LLM to further improve inference-time performance.