aiqbench analysis
This research introduces DLM-AN, a novel system using masked discrete diffusion to enable precise control over accent strength in speech normalization tasks like dubbing and language learning. By selectively reusing source tokens that encode native pronunciation, the method achieves superior word error rates while offering interpretable adjustments to accent retention and rhythm. The work represents a significant advancement in controllable speech synthesis, directly addressing a key limitation in existing accent normalization pipelines.
Controllable Accent Normalization via Discrete Diffusion
arXiv:2603.14275v3 Announce Type: replace-cross Abstract: Existing accent normalization methods do not typically offer control over accent strength, yet many applications-such as language learning and dubbing-require tunable accent retention. We propose DLM-AN, a controllable accent normalization system built on masked discrete diffusion over self-supervised speech tokens. A Common Token Predictor identifies source tokens that likely encode native pronunciation; these tokens are selectively reused to initialize the reverse diffusion process. This provides a simple yet effective mechanism for controlling accent strength: reusing more tokens preserves more of the original accent. DLM-AN further incorporates a flow-matching Duration Ratio Predictor that automatically adjusts the total duration to better match the native rhythm. Experiments on multi-accent English data show that DLM-AN achieves the lowest word error rate among all compared systems while delivering competitive accent reduction and smooth, interpretable accent strength control. The implementation is available at https://github.com/P1ping/DLM-AN
Abstract:Existing accent normalization methods do not typically offer control over accent strength, yet many applications-such as language learning and dubbing-require tunable accent retention. We propose DLM-AN, a controllable accent normalization system built on masked discrete diffusion over self-supervised speech tokens. A Common Token Predictor identifies source tokens that likely encode native pronunciation; these tokens are selectively reused to initialize the reverse diffusion process. This provides a simple yet effective mechanism for controlling accent strength: reusing more tokens preserves more of the original accent. DLM-AN further incorporates a flow-matching Duration Ratio Predictor that automatically adjusts the total duration to better match the native rhythm. Experiments on multi-accent English data show that DLM-AN achieves the lowest word error rate among all compared systems while delivering competitive accent reduction and smooth, interpretable accent strength control. The implementation is available at this https URL
Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Sound (cs.SD)
arXiv:2603.14275 [eess.AS]
(or arXiv:2603.14275v3 [eess.AS] for this version)
https://doi.org/10.48550/arXiv.2603.14275 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 · 239 words
Related coverage
Measurement incompatibility in Bayesian multiparameter quantum estimation
Quantum Journal · Aug 13, 2026
Latent reasoning architectures would likely undermine CoT, our strongest oversight tool
AI Alignment Forum · Sep 23, 2026
Mean-to-Score Discrete Diffusion: Posterior-Mean Denoisers for Score Entropy
arXiv · Jul 24, 2026
When Genomic Masking Priors Fail to Transfer: Strong Variant Prediction, Weak Functional Generation
arXiv · Sep 7, 2026