PolyGlotFake

PolyGlotFake是一个创新的跨语言和多模态深度伪造数据集，由九州大学创建。该数据集包含7种语言的内容，使用了多种先进的文本到语音、声音克隆和唇同步技术。数据集的创建过程涉及从公开视频平台收集高质量视频，并将其内容翻译成其他六种语言，同时使用五种高级声音克隆和TTS技术生成目标语言的音频。PolyGlotFake数据集不仅在技术上具有先进性，还广泛应用于多模态深度伪造检测研究，旨在解决当前深度伪造技术带来的信息安全问题。

PolyGlotFake is an innovative cross-lingual and multimodal deepfake dataset developed by Kyushu University. This dataset contains content in 7 languages, adopting multiple advanced text-to-speech (TTS), voice cloning and lip-sync technologies. The dataset creation process involves collecting high-quality videos from public video platforms, translating their content into the other six languages, and generating audio in target languages using five advanced voice cloning and TTS technologies. Not only technically advanced, the PolyGlotFake dataset is also widely used in multimodal deepfake detection research, aiming to address the information security issues caused by current deepfake technologies.