README.md

---
license: odc-by
task_categories:
- text-classification
- text-generation
- text2text-generation
language:
- ja
tags:
- subtitles
pretty_name: Japanese AnimeSubtitles
---

![image/png](https://cdn-uploads.huggingface.co/production/uploads/633e85093a17ab61de8d9073/ytJWIm9jDtZIYr7pO7pxz.png)

# KaraKaraWitch/AnimuSubtitle-JP

This dataset is an extract of Subtitles from anime. The original files are sourced from nyaa.si.

## Folders

- `data` (Deprecated)
- `data_ass` (Extracted .ass subtitles from \<SOURCE A\>)
- `data_TS` (Extracted arib subtitles from \<SOURCE B\>)

## Dataset Format

Dataset is in Advanced SubStation Alpha (colloquially known as as ASS) [SSA/ASS Specs](https://web.archive.org/web/20170625033533/http://www.tcax.org/docs/ass-specs.htm).

If you're looking for programmatic access, you may parse the file with this [python library](https://pypi.org/project/ass/).

```py
import ass
with open("tests/test.ass", encoding='utf_8_sig') as f:
    doc = ass.parse(f)
    # events are subtitles lines.
    # doc.events 
```

You may open it in a suitable subtitle editing program such as [Aegisub](https://github.com/arch1t3cht/Aegisub/releases/latest).

## License

Tentatively `ODC-By` but this might change since the actual subtitles are from anime which they do have a financial incentive.