VC-CL3
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/coqui-ai/TTS
下载链接
链接失效反馈官方服务：
资源简介：
该数据集旨在探究所提方法在低资源语言环境下的有效性，包含五种语言：加泰罗尼亚语、豪萨语、印度尼西亚语、马来语和泰米尔语。数据来源于FLEURS数据集的真实语音数据，每段真实语音仅被伪造一次。整个数据集跨越12种语言，包含大约120万个样本，但具体到VC-CL3数据集的规模未予提供。该任务专注于低资源语言的语音转换。
This dataset is designed to evaluate the effectiveness of the proposed method in low-resource language scenarios, covering five languages: Catalan, Hausa, Indonesian, Malay, and Tamil. All real speech data is sourced from the FLEURS dataset, with each real speech utterance being forged exactly once. The overall dataset spans 12 languages and contains approximately 1.2 million samples, while the exact scale of the VC-CL3 dataset is not specified. This task focuses on speech conversion for low-resource languages.
提供机构：
Coqui-TTS toolkit
