OnlineGame_Dataset

## 0x0 使用协议：
禁止商用

## 0x1 数据说明：
标注说明：标注，说话人和对应的音频是直接读游戏引擎的脚本生成的，应该是100%准确率，全部存放在index.json里面，如果还有错误可以在开issues反馈（有些遗漏的控制符可能没洗干净）。
务必根据index.json里面的键值对找音频，不在index内的音频请直接丢弃，说话人为？？？，？？等的请直接丢弃。

数据语言：简体中文，日本語，English，한국어

## 0x2 其他：
由于.hca音频文件为CRIWARE的高压缩有损编码格式，解码后二次压缩会造成损耗，因此仓库只提供解密后的hca

如何解析acb，awb容器，如何解密hca，请参考这个仓库：https://github.com/bfloat16/PyCriCodecs

解码hca请使用[pyav](https://github.com/PyAV-Org/PyAV)（libavcodec的python绑定）

0x0 Usage Agreement:
Commercial use is prohibited.

0x1 Data Description:
Annotation Notes: The annotations, speakers and corresponding audio files are directly generated by reading game engine scripts, with an accuracy of 100%. All valid data is stored in index.json. You can submit issues to report any remaining errors (some missing control characters may not have been fully cleaned up). Please locate audio files strictly based on the key-value pairs in index.json. Discard any audio files not listed in index.json, as well as audio files with speakers marked as ??? or similar placeholders.

Data Languages: Simplified Chinese, Japanese, English, Korean.

0x2 Additional Notes:
Since .hca audio files are a high-compression lossy coding format developed by CRIWARE, secondary compression after decoding will cause quality loss. Therefore, this repository only provides decrypted .hca files. For instructions on parsing ACB and AWB containers and decrypting HCA files, please refer to this repository: https://github.com/bfloat16/PyCriCodecs. To decode HCA files, please use [pyav](https://github.com/PyAV-Org/PyAV) (the Python binding for libavcodec).