mlsanityData to 19 Oct 2026

Training-Free Contextual ASR via SpeechLLM-Based Error-Aware Selective Retrieval

arXiv 2609.30694 ↗First posted 25 Sep 2026cs.SD eess.AScited by 0

On the map: Speech and audio language models → Speech recognition models and training

Abstract (arXiv)

Recognition of domain-specific and low-frequency terms remains challenging for automatic speech recognition (ASR). Although contextual biasing can improve their recognition, directly providing a large terminology dictionary introduces many irrelevant biasing terms. Retrieval-based contextual biasing addresses this issue by selecting candidate terms from an external dictionary, but querying many recognized words requires numerous dictionary lookups and may yield poorly targeted candidates. We propose a training-free contextual ASR framework in which a pretrained speech large language model (SpeechLLM) jointly generates an ASR hypothesis and localizes error spans likely to involve domain-specific terms. Only the localized spans are used to retrieve phonologically similar terms from an external terminology dictionary. The same SpeechLLM then re-recognizes the audio conditioned on the first-pass hypothesis and the retrieved terms, without task-specific model training. To assess applicability across domains, we evaluate the framework on medical, air traffic control, and financial speech. The proposed method substantially reduces dictionary queries while improving the recall and ranking of relevant terminology candidates and second-pass ASR performance across all three domains.

Key facts. No citing papers in the map yet (data through 19 Oct 2026). Placed in Speech recognition models and training, one of 11 directions in Speech and audio language models. Builds on 26 of its 35 references that are in the map. First posted on arXiv 25 Sep 2026.

Builds on

26 of 35 references.

Closest ancestors · 15

Same direction, newest first

  1. 2026Contextual Biasing for ASR in Speech LLM with Common Word Cues and Bias Word Position Prediction14 Apr
  2. 2026Contextual Earnings-22: A Speech Recognition Benchmark with Custom Vocabulary in the Wild28 Mar
  3. 2026CLAR: CIF-Localized Alignment for Retrieval-Augmented Speech LLM-Based Contextual ASR26 Mar
  4. 2026Synthetic Data Domain Adaptation for ASR via LLM-based Text and Phonetic Respelling Augmentation11 Mar
  5. 2025PAC: Pronunciation-Aware Contextualized Large Language Model-based Automatic Speech Recognition16 Sep
  6. 2025Attention2Probability: Attention-Driven Terminology Probability Estimation for Robust Speech-to-Text System26 Aug
  7. 2025PMF-CEC: Phoneme-augmented Multimodal Fusion for Context-aware ASR Error Correction with Error-specific Selective Decoding31 May
  8. 2025BR-ASR: Efficient and Scalable Bias Retrieval Framework for Contextual Biasing ASR in Speech LLM25 May
  9. 2024Contextualization of ASR with LLM using phonetic retrieval-based augmentation11 Sep
  10. 2024Retrieval Augmented Correction of Named Entity Speech Recognition Errors9 Sep
  11. 2024Improving Neural Biasing for Contextual Speech Recognition by Early Context Injection and Text Perturbation14 Jul
  12. 2023ed-cec: improving rare word recognition using asr postprocessing based on error detection and context-aware error correction8 Oct
3 more
  1. 2022RED-ACE: Robust Error Detection for ASR using Confidence Embeddings14 Mar
  2. 2021Tree-constrained Pointer Generator for End-to-end Contextual Speech Recognition1 Sep
  3. 2021Contextualized Streaming End-to-End Speech Recognition with Trie-Based Deep Biasing and Shallow Fusion5 Apr

Other references · 11

  1. 2017Attention Is All You Needcited by 61,180
  2. 2019Common Voice: A Massively-Multilingual Speech Corpuscited by 1,596
  3. 2025Qwen3-Omni Technical Reportcited by 578
  4. 2021GigaSpeech: An Evolving, Multi-domain ASR Corpus with 10,000 Hours of Transcribed Audiocited by 497
  5. 2018Deep context: end-to-end contextual speech recognitioncited by 152
  6. 2022Earnings-22: A Practical Benchmark for Accents in the Wildcited by 65
  7. 2023COSMIC: Data Efficient Instruction-tuning For Speech In-Context Learningcited by 38
  8. 2021An evaluation of word-level confidence estimation for end-to-end automatic speech recognitioncited by 22
  9. 2021Multi-Task Learning for End-to-End ASR Word and Utterance Confidence with Deletion Predictioncited by 11
  10. 2020Acoustic Neighbor Embeddingscited by 7
  11. 2024A Theoretical Framework for Acoustic Neighbor Embeddingscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Contextual Biasing for LLM-Based ASR with Hotword Retrieval and Reinforcement Learning
  2. 2025Contextualized Automatic Speech Recognition with Dynamic Vocabulary Prediction and Activation
  3. 2026BaLEEN: Biasing with Latent Encoded Entities for Context-Aware ASR
  4. 2024Locality enhanced dynamic biasing and sampling strategies for contextual ASR
  5. 2026How to Recognize New Words: A Comparison Between Context Biasing Methods and Speech LLMs
  6. 2022Minimising Biasing Word Errors for Contextual ASR with the Tree-Constrained Pointer Generator
  7. 2025ContextASR-Bench: A Massive Contextual Speech Recognition Benchmark
  8. 2025A Neural Model for Contextual Biasing Score Learning and Filtering