czyang/MultiFoley-VGGSound-Test-Audio

这个数据集包含了基于MultiFoley工作的筛选后的VGGSound测试用例生成的结果。对于每个8秒的视频，生成了4个样本，这些样本是通过无声视频和文本（简化为VGGSound类别名称）输入生成的。每个音频文件按照特定的格式命名，包括类别名称、唯一标识符、开始时间和索引。整个数据集的大小约为25GB。

This dataset contains the generated results of our MultiFoley work based on the filtered VGGSound test cases. For each 8-second video, 4 samples are generated, which are produced from silent video inputs and text inputs (simplified to VGGSound category names). Each audio file is named in a specific format, including the category name, unique identifier, start time, and index. The entire dataset is approximately 25GB in size.