AI Voice Chat in Unreal Engine
Players are used to talking to AI assistants in their daily lives. Now that same experience can happen inside your game. Real-time voice AI lets players speak to NPCs naturally and hear spoken responses — no pre-recorded dialogue, no text boxes, just conversation.
The Voice AI Stack
A complete voice AI pipeline has three parts:
- Speech-to-text (STT) — convert the player's spoken words into text
- LLM processing — generate an intelligent response
- Text-to-speech (TTS) — convert the response back into spoken audio
Or, with realtime voice APIs, all three happen in a single round trip.
Available Providers
OpenAI Realtime API
Voice-to-voice in a single WebSocket connection. The player speaks, GPT-4o processes, and a voice responds — all in real-time with sub-second latency. The most seamless voice AI experience available.
Google Gemini Live
Google's equivalent to OpenAI Realtime. Real-time multimodal conversations with Gemini, supporting voice input and output.
ElevenLabs
The gold standard for AI voice quality. Dozens of preset voices plus custom voice cloning. Create unique character voices that sound natural and expressive. Also offers speech-to-text via Scribe.
OpenAI TTS
Six built-in voices with fast generation. Simpler than ElevenLabs but integrated with the same API as ChatGPT.
Google Cloud TTS
Hundreds of voices across dozens of languages. WaveNet and Neural2 voices for natural-sounding speech.
OpenAI Whisper
Industry-leading speech-to-text. Handles accents, background noise, and multiple languages. Convert player speech to text for LLM processing.
Integration
GenAI for Unreal supports the complete voice AI stack — all providers listed above work through Blueprint and C++ APIs with streaming audio playback.
Use Cases
- Voice-interactive NPCs — players talk, characters respond with unique voices
- AI narrators — dynamic storytelling with professional voice quality
- Accessibility — voice input for players who can't type, voice output for players who can't read
- Language learning games — practice conversation with AI characters in any language
- VR interactions — voice is the most natural input method in VR