Groove

该数据集是专为Groove训练而设计的，包含了大约8小时的音频资料。这些资料被用于训练模型，以评估回声水印技术。其规模大约为8小时，任务涵盖了音频生成和回声水印评估两个方面。

This dataset is specifically designed for Groove training, containing approximately 8 hours of audio data. These audio materials are used to train models for evaluating echo watermarking technology. With a total duration of around 8 hours, its tasks cover two aspects: audio generation and echo watermarking evaluation.