LiveVoice AI Pro - Full Voice-to-Voice AI Suite - Compatible with Gemini Live

LiveVoice AI Pro
Bring your project to life with the most advanced, real-time multilingual AI integration available for Unity. The Pro edition unlocks the full potential of conversational agents with Google’s Gemini Live API, enabling true conversational AI with bidirectional voice streaming, tool execution, and internet access.
Why use this integration?
True Voice-to-Voice: Includes a robust Microphone implementation with Voice Activity Detection (VAD) for natural, hands-free conversation.
Game Control (Function Calling): Connect the AI directly to your C# methods. Let players spawn items, open doors, or control UI just by speaking.
Unbeatable Speed & Cost: Bypasses the lag and high cost of chaining multiple services (STT + LLM + TTS). LiveVoice AI does it all in one low-latency WebSocket stream.
Smart & Connected: Enable Google Search for up-to-date real-world info and Thinking Mode for complex logic.
Extensive Documentation: All features are properly documented, including samples on how to use all functionality.
Fully Multilingual: LiveVoice AI supports 29+ languages natively, allowing you to create for every use case.
Battle Tested: This integration has been tested extensively in educational projects running Unity 6 & 6.3LTS

PRO Exclusive Features
Microphone Input & VAD: Built-in, configurable Voice Activity Detection detects when the player speaks and streams audio automatically.
AI ↔ C# Function Calling: Define C# functions that the AI can trigger based on conversation context (e.g., LightTorch(), OpenInventory()).
Google Search Access: Give your AI the ability to search the web for real-time answers.
Configurable Thinking Mode: Enable deeper reasoning capabilities for puzzle-solving or complex narrative NPCs.
30+ Premium Voices: Access the complete library of high-quality Google AI voices.
Input Transcription: Receive text transcripts of what the user said (Speech-to-Text), not just the AI's response.

Core Features (Included)
Seamless WebSocket Streaming: High-performance, asynchronous communication.
Full Context Control: Sliding context window to manage memory usage.
Custom System Instructions: Define the exact persona, lore, and behavior of your AI.
Output Control: Fine-tune Temperature, Top-P, Top-K, and Max Tokens.
Sanitization: Basic and extendable text cleaning for safe outputs.

Robust Event System
Build complex interactions with a complete suite of real-time C# events. Here are some examples, with a lit more events available for your projects:
GeminiTranscribeEvent: Get instant text transcripts as the user speaks or as the AI responds.
GeminiTurnEndEvent: Fires the moment a full conversational "turn" is finished.
GeminiGenerationCompleteEvent: Signals that the AI has finished generating its response.
GeminiServerClosedSessionEvent: This event provides specific error codes and closure reasons (like network timeouts or API limits), allowing you to implement seamless reconnection logic.
GeminiSessionReadyEvent: Know exactly when the WebSocket handshake is complete and the AI is "listening"

Easy to get started
Get a Key: Visit Google AI Studio to acquire your API key.
Import: Add the package to your project.
Run: Open the Sample Scene, paste your key, and start talking immediately.
Extend: Use the included scripts to hook up Function Calling and VAD to your gameplay logic.

API Usage
You are required to use your own API key in order to use this tool. Gemini Live currently can be used for free with data sharing, or paid without data sharing. Reference the Gemini Live API Pricing for more details.
Threre is an easy method to provide your key in a scriptable object, which can be added to your gitignore to keep keys private.
Disclaimer This project is not affiliated, associated, authorized, endorsed by, or in any way officially connected with Google LLC. It is an independent integration providing a Unity-compatible interface for the public Gemini Live API.