EmoFake
收藏资源简介:
EmoFake数据集是中国科学院自动化研究所模式识别国家重点实验室开发的,专注于情感伪造音频检测。该数据集包含40900条音频数据,由五种情感状态(中性、快乐、愤怒、悲伤和惊讶)的英语语音组成,通过七种开源情感语音转换模型生成伪造音频。数据集的创建旨在通过提供多样化的情感转换音频,推动情感伪造音频检测技术的发展,特别是在智能设备交互和个性化语音生成领域。
EmoFake Dataset was developed by the State Key Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, focusing on emotional fake audio detection. This dataset contains 40,900 audio samples, consisting of English speech in five emotional states: neutral, happy, angry, sad and surprised. The fake audio samples are generated via seven open-source emotional voice conversion models. The development of this dataset aims to advance the development of emotional fake audio detection technologies by providing diverse emotionally converted audio data, particularly in the fields of smart device interaction and personalized speech generation.




