> For the complete documentation index, see [llms.txt](https://benjaminx.gitbook.io/projectatri/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://benjaminx.gitbook.io/projectatri/technical-overview-ji-shu-gai-lan/models-mo-xing.md).

# Models 模型

* Large language model: from ChatGPT

* Speech inference: client-side VITS - ncnn

* Graphics rendering: Native Live2D

* Speech recognition: client-side Sherpa - ncnn

* UI framework: Jetpack Compose

* Project structure: Retrofit + MVVM + LiveData/Flow + Room

* Multiple ncnn libraries seem to be process-exclusive resources, with AIDL used for inter-process communication

* Lip-sync: integrated Native version of meta-lipSync

* Integrated Baidu translation; if needed, set appid and key yourself. Can be turned off in the settings page.

* 语言大模型来自 GhatGPT

* 语音推理为客户端本地 VITS - ncnn

* 图形渲染基于 Native Live2D

* 语音输入识别为客户端本地 Sherpa - ncnn

* UI 框架为 Jetpack Compose

* 项目结构为 Retrofit + MVVM + LiveData/Flow + Room

* 多个 ncnn 库似乎是进程独享资源，使用 AIDL 做进程间通信

* 嘴形同步接入了 Native 版 meta - lipSync

* 接了 baidu 翻译功能，如需则自行设置 appid 和 key，不需要可在 setting 页面关闭翻译功能