SFR
收藏arXiv2023-09-15 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2208.10489v3
下载链接
链接失效反馈官方服务：
资源简介：
SFR数据集是由中国科学院自动化研究所创建，专注于深度伪造音频的系统指纹识别。该数据集包含181,764条由七家中国主流语音合成系统生成的音频，分为清晰和压缩两种格式，以模拟真实世界中的数据传播情况。数据集旨在支持模型版权保护和数字证据法医分析等应用，解决识别音频来源的问题。
The SFR Dataset was created by the Institute of Automation, Chinese Academy of Sciences, focusing on system fingerprint recognition for deepfake audio. This dataset contains 181,764 audio samples generated by seven mainstream Chinese speech synthesis systems, and is provided in both uncompressed and compressed formats to simulate real-world data propagation scenarios. The dataset is intended to support applications such as model copyright protection and digital evidence forensic analysis, addressing the problem of identifying the source of audio.
提供机构：
中国科学院自动化研究所
创建时间：
2022-08-21
