aiqbench analysis
This research introduces supervised contrastive learning as a lightweight regularization technique to enhance automatic speech recognition (ASR) systems against accent variability without altering model architecture. By stabilizing encoder representations, the method achieves significant word error rate reductions on unseen accents across multiple pre-trained models, offering an effective strategy for building more inclusive ASR technologies.
Contrastive Regularization for Accent-Robust ASR
arXiv:2605.03297v2 Announce Type: replace-cross Abstract: ASR systems based on self-supervised acoustic pretraining and CTC fine-tuning achieve strong performance on native speech but remain sensitive to accent variability. We investigate supervised contrastive learning (SupCon) as a lightweight, accent-invariant auxiliary objective for CTC fine-tuning. An utterance-level contrastive loss regularizes encoder representations without architectural modification or explicit accent supervision. Experiments on the L2-ARCTIC benchmark show consistent WER reductions across multiple pretrained encoders, with up to 25 -- 29\% relative reduction under unseen-accent evaluation. Analysis using within-transcript cosine dispersion indicates that SupCon promotes more compact and stable representation geometry under accent variability. Overall, SupCon provides an effective and model-agnostic regularization strategy for improving accent robustness.
Abstract:ASR systems based on self-supervised acoustic pretraining and CTC fine-tuning achieve strong performance on native speech but remain sensitive to accent variability. We investigate supervised contrastive learning (SupCon) as a lightweight, accent-invariant auxiliary objective for CTC fine-tuning. An utterance-level contrastive loss regularizes encoder representations without architectural modification or explicit accent supervision. Experiments on the L2-ARCTIC benchmark show consistent WER reductions across multiple pretrained encoders, with up to 25 -- 29\% relative reduction under unseen-accent evaluation. Analysis using within-transcript cosine dispersion indicates that SupCon promotes more compact and stable representation geometry under accent variability. Overall, SupCon provides an effective and model-agnostic regularization strategy for improving accent robustness.
Sound (cs.SD); Machine Learning (cs.LG)
arXiv:2605.03297 [cs.SD]
(or arXiv:2605.03297v2 [cs.SD] for this version)
https://doi.org/10.48550/arXiv.2605.03297 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 200 words
Related coverage
Type-IV Code Clone Detection via Layer-Wise Non-Contrastive Representation Learning
arXiv · Sep 16, 2026
Dynamic Gated Cross-Modal Fusion with Sarcastic-aware Contrastive Regularization for Multimodal Sarcasm Detection
arXiv · Aug 21, 2026
SAFE-MEME: Structured Reasoning Framework for Robust Hate Speech Detection in Memes
arXiv · Jul 28, 2026
IConE: Batch Independent Collapse Prevention for Self-Supervised Representation Learning
arXiv · Jul 23, 2026