sound_events
收藏资源简介:
Sound Events Dataset 是一个包含64个音频文件的数据集,涵盖了23种不同的声音事件类别。这些音频文件以WAV/MP3/OGG格式录制,并转换为Parquet格式以便于Hugging Face Datasets使用。数据集中的声音事件类别包括搜索木制抽屉、拉动木椅、开关笔、开关灯、开门关门、行走、坐在木椅上、键盘打字等,每种声音事件都有详细的描述。数据集结构上,音频文件按类别存放在`sound_audios/`文件夹的子目录中,并附带一个`metadata.csv`文件,提供音频文件路径、标签和描述信息。该数据集适用于声音事件识别、音频分类等任务。
Sound Events Dataset is a dataset containing 64 audio files spanning 23 distinct sound event categories. These audio files are recorded in WAV, MP3 and OGG formats, and converted to the Parquet format for convenient use with Hugging Face Datasets. The sound event categories covered in this dataset include searching wooden drawers, pulling wooden chairs, opening and closing pens, toggling lights, opening and closing doors, walking, sitting on wooden chairs, keyboard typing and more, with detailed descriptions provided for each sound event. In terms of dataset structure, audio files are sorted into subdirectories under the `sound_audios/` folder by category, and are accompanied by a `metadata.csv` file that provides audio file paths, labels and descriptive information. This dataset is applicable to tasks such as sound event recognition and audio classification.




