laion/voices-with-captions

合成语音描述数据集包含3180个合成的语音样本，每个样本都配有一个简短的描述，描述内容大致包括声音的年龄、性别、口音以及一些语音的一般特征（例如“老妇人，爱尔兰口音”）。所有语音均为合成生成，不代表任何真实个人，适合用于训练和评估语音转换模型，而不会侵犯个人身份或权利。

This dataset contains 3,180 synthetically generated voice samples, each paired with a short caption describing the voice, including age, gender, accent, and sometimes general vocal qualities (e.g., old woman, Irish accent). All voices are synthetically generated and do not represent any real individuals, making the dataset suitable for training and evaluating voice conversion models without infringing on personhood or identity rights.