超越伪影：基于音乐内在特征的可泛化合成歌曲检测
Beyond Artifacts: Towards Generalizable Synthetic Song Detection via Music-Intrinsic Features
AI总结：
提出Sofia框架，通过特征特定专家和自适应混合专家模型利用音乐内在特征（人声、音频效果、全局结构）进行合成歌曲检测，在MUSIC8K基准上F1提升18.5点，具有强鲁棒性。
摘要：
AI音乐生成器的快速发展凸显了对可靠合成歌曲检测（SSD）的迫切需求。现有SSD方法通常依赖于低级伪影或固定特征假设，难以捕捉生成器无关的线索。为解决这一问题，我们提出Sofia（基于音乐特征的合成歌曲检测框架），一个灵活的框架，通过特征特定专家和自适应混合专家（MoE）模块对音乐内在属性进行建模。通过使用代表性的人声、音频效果、全局结构特征及其组合配置Sofia，我们展示了它们的个体和互补贡献。为全面评估我们的框架，我们进一步构建了MUSIC8K，一个具有挑战性的基准，包含最新出现的生成器和逼真的音频扰动。实验表明，Sofia从音乐内在特征中学习生成器无关的表示，在MUSIC8K-O上相比最强基线F1分数提升18.5点，同时保持强鲁棒性。
英文摘要：
The rapid advancement of AI music generators highlights the urgent need for reliable Synthetic Song Detection (SSD). Existing SSD methods often rely on low-level artifacts or fixed feature assumptions, struggling to capture generator-agnostic cues. To address this, we propose Sofia (Synthetic-song detection framework via music features), a flexible framework that models music-intrinsic attributes via feature-specific experts and an adaptive Mixture-of-Experts (MoE) module. By configuring Sofia with representative Vocal, Audio-effect, Global structure features, and their combinations, we present their individual and complementary contributions. To comprehensively evaluate our framework, we further construct MUSIC8K, a challenging benchmark featuring lastest emerging generators and realistic audio perturbations. Experiments show that Sofia learns generator-agnostic representations from music-intrinsic features, improving the F1 score by 18.5 points over the strongest baseline on MUSIC8K-O while maintaining strong robustness.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
