mlsanityData to 14 Oct 2026

MGSC: A Multi-granularity Consistency Framework for Robust End-to-end Asr

arXiv 2508.15853 ↗First posted 20 Aug 2025cs.CL cs.AI cs.SDcited by 0

Not on the map. Why?

Abstract (arXiv)

End-to-end ASR models, despite their success on benchmarks, often pro-duce catastrophic semantic errors in noisy environments. We attribute this fragility to the prevailing 'direct mapping' objective, which solely penalizes final output errors while leaving the model's internal computational pro-cess unconstrained. To address this, we introduce the Multi-Granularity Soft Consistency (MGSC) framework, a model-agnostic, plug-and-play module that enforces internal self-consistency by simultaneously regulariz-ing macro-level sentence semantics and micro-level token alignment. Cru-cially, our work is the first to uncover a powerful synergy between these two consistency granularities: their joint optimization yields robustness gains that significantly surpass the sum of their individual contributions. On a public dataset, MGSC reduces the average Character Error Rate by a relative 8.7% across diverse noise conditions, primarily by preventing se-vere meaning-altering mistakes. Our work demonstrates that enforcing in-ternal consistency is a crucial step towards building more robust and trust-worthy AI.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 19 of its 26 references that are in the map. First posted on arXiv 20 Aug 2025.

Builds on

19 of 26 references.

References · 19

  1. 2022Robust Speech Recognition via Large-Scale Weak Supervisioncited by 4,057
  2. 2019SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognitioncited by 1,380
  3. 2017AISHELL-1: An Open-Source Mandarin Speech Corpus and A Speech Recognition Baselinecited by 373
  4. 2023Matcha-TTS: A fast TTS architecture with conditional flow matchingcited by 145
  5. 2024Automatic Speech Recognition using Advanced Deep Learning Approaches: A surveycited by 50
  6. 2022CLNode: Curriculum Learning for Node Classificationcited by 24
  7. 2024Improving Robustness of LLM-based Speech Synthesis by Learning Monotonic Alignmentcited by 22
  8. 2024CR-CTC: Consistency regularization on CTC for improved speech recognitioncited by 17
  9. 2024Improving Neural Biasing for Contextual Speech Recognition by Early Context Injection and Text Perturbationcited by 11
  10. 2024Contextualized Automatic Speech Recognition with Dynamic Vocabularycited by 8
  11. 2022Self-Remixing: Unsupervised Speech Separation via Separation and Remixingcited by 5
  12. 2022Improving CTC-based ASR Models with Gated Interlayer Collaborationcited by 4
7 more
  1. 2022Supervised Attention in Sequence-to-Sequence Models for Speech Recognitioncited by 3
  2. 2023Unsupervised Improvement of Audio-Text Cross-Modal Representationscited by 3
  3. 2024Multi-Task Learning Using Uncertainty to Weigh Losses for Heterogeneous Face Attribute Estimationcited by 2
  4. 2025Boosting CTC-Based ASR Using LLM-Based Intermediate Loss Regularizationcited by 2
  5. 2025CitySeg: A 3D Open Vocabulary Semantic Segmentation Foundation Model in City-scale Scenarioscited by 2
  6. 2023Factual Consistency Oriented Speech Recognitioncited by 1
  7. 2023Improved Contextual Recognition In Automatic Speech Recognition Systems By Semantic Lattice Rescoringcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025On Almost Surely Safe Alignment of Large Language Models at Inference-Time
  2. 2026Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization
  3. 2026Trust Your Guide Only When Certain: Uncertainty-Aware Sparse Alignment at Inference Time
  4. 2026Better Retrieval, Worse Robustness: How Multi-hop RAG Amplifies Upstream ASR Errors
  5. 2026Beyond Position Bias: Shifting Context Compression from Position-Driven to Semantic-Driven
  6. 2026Reliable Neural-Codec Text-to-Speech by ASR Self-Verification and Distillation: Near-Zero Catastrophic Failures Across Models and Codecs
  7. 2026WiSparse: Boosting LLM Inference Efficiency with Weight-Aware Mixed Activation Sparsity
  8. 2025Trusted Uncertainty in Large Language Models: A Unified Framework for Confidence Calibration and Risk-Controlled Refusal