Sample scenes demonstrating different OpenAI real-time conversation capabilities. Clone the full repository to access these samples.
Location: Assets/Samples/OpenAI/RealtimeConvSample/
Basic real-time conversation implementation with visual feedback.
OpenAI.Demo.mp4
Features:
- Speech-to-speech conversation
- 3D orb visualization
- Configurable AI prompts
Setup:
- Open
RealtimeConvSample.unity - Configure OpenAI settings on RealtimeConversationManager
- Run and start speaking
Location: Assets/Samples/OpenAI/SimpleVoiceActionSample/
Voice-controlled actions with function calling.
SimpleAgenticToolDemo.mov
Features:
- Voice commands for object manipulation
- Custom tool integration
- JSON schema-based actions
How it works:
VoiceActionTool.csimplements methods that match the AI function callsActionsSchema.jsondefines the available tools for the AI- Unity Events connect AI calls to scene actions
Setup:
- Open
SimpleVoiceActionsSample.unity - Configure OpenAI settings
- Try voice commands like "can you make sping the cube?"
Location: Assets/Samples/OpenAI/PCAConversationalAISample/
Conversational AI with visual input capabilities.
PCA.Demo.mp4
Features:
- Camera integration
- Image-based conversations
- Real-time visual processing
Setup:
- Open
PCAConversationalAISample.unity - Configure OpenAI settings
- Point camera at objects and ask the AI about what it sees
Location: Assets/Samples/OpenAI/AgenticToolsSample/
Advanced tool integration with mesh and line art generation.
Line.Art.Tool.Demo.mp4
Scenes:
LineArtToolsSample.unity: AI-generated line artMeshToolsSample.unity: 3D mesh generation tools
Features:
- AI-controlled content generation
- Advanced tool integration
- Creative AI applications
Setup:
- Open desired sample scene
- Configure OpenAI settings
- Ask the AI to generate content based on your descriptions
Additional Setup for MeshToolsSample:
For the MeshToolsSample.unity scene, you need to run the Python mesh tool server locally:
- Prerequisites: Ensure you have Python 3.11 installed
- Navigate to MeshTool folder: Open terminal and go to
MeshTool/directory - Install dependencies:
pip install -r requirements.txt
- Run the mesh tool server:
python mesh_tool.py
- Verify server is running: You should see "Server started successfully!" on port 8765
- Now run the Unity scene: The MeshToolsSample will connect to the local Python server
The Python server provides advanced 3D mesh generation capabilities that Unity calls through WebSocket communication.
- Prerequisites: Ensure you have an OpenAI API key
- Open Scene: Navigate to the sample scene you want to try
- Configure: Set up your OpenAI config asset in the scene
- Test: Press Play and interact with voice