用作者身份验证理论评估LLM风格个性化，揭示模型模仿个人文风时的真实差距与评测盲点。
arXiv:2604.26460v2 Announce Type: replace Abstract: Stylistic personalization - making LLMs write in a specific individual's style, rather than merely…
把风格训练压缩到单次模型前向，扩散迁移效率与保真度的新平衡方案
arXiv:2606.13809v1 Announce Type: new Abstract: Diffusion-based style transfer must balance inference efficiency with stylization fidelity. Adapter-ba…
颠覆传统认知：语音克隆本质是风格迁移，斯坦福团队用新视角重新定义技术边界。
arXiv:2605.16578v1 Announce Type: cross Abstract: Artificially generated speech is increasingly embedded in everyday life. Voice cloning in particular…