VocalAuth40
收藏IEEE2026-04-17 收录
下载链接：
https://ieee-dataport.org/documents/vocalauth40
下载链接
链接失效反馈官方服务：
资源简介：
The paper introduces a custom dataset, called VocalAuth40, containing 40 audio samples in the English language, obtained from various YouTube videos. These audios are used to evaluate model performance on real-world samples, to check the generalization ability of the model. The data contains audios of duration 10-20 seconds, in .mp3 and .wav format, with 20 spoofed and 20 real audio samples, equally divided among males and females. The real audio clips collected include speeches by politicians, celebrity interviews, and clips from TV Shows. The spoofed audios are both impersonations of real-world figures and AI-generated voice-overs.
提供机构：
Shweta Jindal; Tanya Gupta; Pallika Dhingra; Vaishnavi Saxena; Saachi Bansal
