arXiv cs.CL· Kehinde Elelu, Joshua E. Siegel, Mohammadali Saffary, Tashfain Ahmed, Simeon Babatunde, Ebuka Okpala·· 13 小时前AI 评分24
ConformalMOS：用共形区间与序数建模实现不确定性感知的 MOS 预测
ConformalMOS: Uncertainty-Aware MOS Prediction with Conformal Intervals and Ordinal Modeling
AI 导读
ConformalMOS 框架在 MOS 预测中引入共形预测，在可交换性假设下给出具有覆盖保证的统计有效预测区间，同时保留点估计。训练时将 one-hot 目标转为高斯平滑分布以保留分数顺序，改善了 softmax 置信度过高时的校准。在 BVCC 数据集上 MSE 为 0.08，并经验证了区间覆盖率。
正文 · 原文
Abstract:Accurate prediction of human Mean Opinion Scores (MOS) is crucial for evaluating synthetic speech quality in text-to-speech (TTS) and voice conversion (VC). Most MOS predictors output only point estimates and ignore uncertainty, limiting reliability in model selection and deployment. We propose ConformalMOS, a framework that augments MOS prediction with conformal prediction to produce statistically valid prediction intervals with guaranteed coverage under exchangeability, alongside point estimates. During training, MOS labels are modeled in an ordinal-aware manner by converting one-hot targets into Gaussian-smoothed distributions, preserving score ordering. This improves uncertainty calibration, particularly when softmax confidence is overconfident on out-of-distribution speech. On BVCC datasets, ConformalMOS achieves competitive accuracy (MSE = 0.08) while delivering empirically validated coverage, enhancing reliability for real-world TTS and VC deployment.
| Subjects: | Sound (cs.SD); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.36126 [cs.SD] |
| (or arXiv:2610.36126v1 [cs.SD] for this version) | |
| https://doi.org/10.48550/arXiv.2610.36126 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tashfain Ahmed [view email]
[v1]
Mon, 26 Oct 2026 18:35:12 UTC (279 KB)
来源：arXiv cs.CL · arxiv.org