ocisd4/superb-train

该数据集包含三个不同的配置（ic、ks、sd），每个配置都包含音频和消息两个特征。音频的采样率为16000Hz，消息包含内容和角色两个字段。每个配置都有一个训练集分割，分别提供了字节大小和示例数量。此外，还提供了下载大小和数据集大小的信息。

The dataset contains three different configurations (ic, ks, sd), each with two features: audio and messages. The audio has a sampling rate of 16000Hz, and the messages include content and role fields. Each configuration has a training split, providing the number of bytes and examples. Additionally, the download size and dataset size are provided.