MultimodalSensor / Time-seriesVideo
Open X-Embodiment
1M+ real-robot trajectories from 20+ institutions.
1M–10M·Apache-2.0·rosbag
AudioMultimodalText
SoulTide-AudioData-Dataset
目录结构 character/ └── char / ├── resource/ │ ├── audio/ 原始音频 │ ├── srt/ 原始srt │ └── processed/ 利用 Process-Resource 根据原始资源处理后的资源 ├── recognized/…
目录结构 character/ └── char / ├── resource/ │ ├── audio/ 原始音频 │ ├── srt/ 原始srt │ └── processed/ 利用 Process-Resource 根据原始资源处理后的资源 ├── recognized/ Whisper-LargeV2 识别的srt ├── calibrated/ 人工校准的srt └── tmp/ build临时文件 工具 搭配此管理器来生成所需的训练集: 搭配此图形工具快速的人工校准任何字幕模型(例Whisper)的生成结果: See the full description on the dataset page:
Source: Hugging Face Hub (zwa73/SoulTide-AudioData-Dataset). Metadata imported from the dataset’s Hub tags.