Genuwin is an Android application that provides an interactive AI companion using Live2D animation. It integrates with Ollama for AI chat, and OpenAI-compatible APIs for Text-to-Speech (TTS) and Speech-to-Text (STT). Your data. Your rules. Genuinely.
- Live2D Animation: Interactive 2D character animation.
- AI Chat: Conversational AI powered by Ollama.
- Voice Interaction: Full audio pipeline with STT and TTS.
- Smart Home Integration: Control Home Assistant devices.
- Advanced Tool System: Weather, search, and more.
- Augmented Reality: AR mode for immersive experiences.
- Android Studio or Gradle
- Java 17
- Live2D Cubism SDK for Java
- An Ollama server and OpenAI-compatible TTS/STT servers (see credits), OR an OpenAI API key
- Home Assistant (Optional)
-
Clone the repository:
git clone https://github.com/shifusen329/Genuwin.git cd Genuwin
-
Setup Live2D SDK:
- Download the Live2D Cubism SDK for Java.
- Extract the downloaded SDK.
- IMPORTANT: Copy the entire
CubismJavaSamples
directory to the parent directory of your Genuwin project.
Your directory structure should look like this:
parent-directory/ ├── Genuwin/ # This repository │ ├── src/ │ ├── build.gradle │ └── settings.gradle └── CubismJavaSamples/ # Live2D SDK (place here) ├── Framework/ │ └── framework/ └── Core/ └── android/ └── Live2DCubismCore.aar
The build system expects
CubismJavaSamples
to be in the parent directory of the Genuwin project, not inside it. -
Configure API Endpoints:
- Copy
src/main/assets/config.properties.template
tosrc/main/assets/config.properties
. - Edit
src/main/assets/config.properties
with your server URLs or your OpenAI API key.
- Copy
-
Build and Install:
- Android Studio: Open the project and build the APK.
- Gradle: Run the following command in the project root:
./gradlew assembleDebug
- Install the generated APK located at
app/build/outputs/apk/debug/app-debug.apk
on your device.
This project is built upon the following open-source software:
- Ollama: Licensed under the MIT License. See https://github.com/ollama/ollama.
- openedai-speech: Licensed under the AGPL-3.0 License. See https://github.com/matatonic/openedai-speech.
- openedai-whisper: Licensed under the AGPL-3.0 License. See https://github.com/matatonic/openedai-whisper.
This project also uses the Live2D Cubism SDK, which has its own licensing terms. Please review the Live2D license before commercial use.