@inproceedings{cao-etal-2026-scores,
    title = "From Scores to Preferences: Redefining Evaluation Paradigm for Speech Quality Reward Modeling",
    author = "Cao, Yifei  and
      Jiang, Changhao  and
      Zhuang, Jiabao  and
      Sun, Jiajun  and
      Zhang, Ming  and
      Xi, Zhiheng  and
      Li, Hui  and
      Dou, Shihan  and
      Wang, Yuran  and
      Zhang, Yunke  and
      Ji, Tao  and
      Gui, Tao  and
      Zhang, Qi  and
      Huang, Xuanjing",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.1638/",
    pages = "32731--32749",
    ISBN = "979-8-89176-395-1",
    abstract = "Speech quality assessment (SQA) is typically formulated as a score regression task based on subjective ratings, such as the Mean Opinion Score (MOS), which inherently suffer from inconsistent standards and limit cross-dataset training and evaluation. To address these limitations, we reformulate SQA as a preference-based comparison paradigm and construct MOS-Pref, a large-scale MOS-derived preference dataset. Building on MOS-Pref, we systematically implement and evaluate three reward modeling paradigms: scalar, semi-scalar, and generative reward models, alongside existing SQA approaches. Our experiments reveal three key findings: (1) scalar models achieve the strongest overall performance, consistently exceeding 74{\%} accuracy; (2) score regression-based approaches generally underperform preference-based methods in both overall performance and generalization; and (3) all reward models struggle on pairs with very small MOS gap. Motivated by these observations, we propose a MOS-aware GRM design that incorporates MOS gap into the reward function during reinforcement learning. Experimental results show that the MOS-aware GRM significantly improves fine-grained speech quality discrimination. We hope this work fosters more rigorous and scalable research in SQA."
}