arXiv · ML· Julius Richter, Yoshiki Masuyama, Christoph Boeddeker, Takahiro Edo, Gordon Wichern, Jonathan Le Roux·· 10 天前AI 评分37
Stochastic Interpolant Prior for Speech (SIPS) 用于语音增强与分离
Predictive-Generative Drift Decomposition for Speech Enhancement and Separation
AI 导读
提出即插即用框架 Stochastic Interpolant Prior for Speech (SIPS),通过分解采样动力学中的预测器诱导漂移和基于分数的去噪分量,增强预训练预测器。
来源:arXiv · ML · arxiv.org