Singing Voice Recordings

该数据集包含了来自12位不同中国歌手的录音，总时长达到59小时，所有录音均具有专业录音室级别的音质。为了增加训练的多样性，该数据集经过了增强处理。此外，该数据集已被用于评估模型在处理未见过的语言和说话者时的性能表现。规模上，数据集涵盖了59小时的声音资料，涉及12位不同的歌手。其任务是对歌唱声音进行合成（Singing Voice Synthesis，简称Svs）。

This dataset contains recordings from 12 distinct Chinese singers, with a total duration of 59 hours, and all recordings feature professional studio-grade audio quality. To enhance training diversity, this dataset has undergone data augmentation. Furthermore, this dataset has been utilized to evaluate model performance when handling unseen languages and speakers. In terms of scale, the dataset encompasses 59 hours of audio materials involving 12 distinct singers. The target task of this dataset is Singing Voice Synthesis (abbreviated as SVS).