Inspiration We were inspired by the gap between how humans experience stories and how AI typically delivers information. Most AI systems talk too much, instruct too clearly, and break immersion. EchoTurn explores a different idea: What if AI guided experiences through rhythm, tone, and story instead of commands?
What it does EchoTurn is a voice-driven AI storytelling engine that listens to natural speech and responds with short narrative echoes—poetic, non-directive fragments designed to guide experience without instructions. Instead of telling users what to do, EchoTurn: Shapes pacing and mood Reinforces emotional context Uses silence and timing as part of interaction It functions as an ambient narrative layer inside immersive applications.
How we built it EchoTurn was built using Unreal Engine 5 as the real-time runtime and interaction layer. The system combines: Unreal Engine 5 for real-time interaction, audio playback, and system orchestration Voice input captured and routed from within Unreal Google Gemini for contextual reasoning and narrative intelligence System-level constraints to ensure AI responses remain non-instructional ElevenLabs for expressive, emotionally rich voice synthesis AI responses are triggered contextually inside Unreal rather than as immediate chatbot replies, enabling natural pacing.
Challenges we ran into Preventing the AI from defaulting to instructional language Maintaining narrative coherence with very short responses Designing pacing where silence feels intentional Integrating AI services into a real-time Unreal Engine workflow
Accomplishments that we're proud of Built a working AI narrative system inside Unreal Engine 5 Created a non-directive AI interaction model Demonstrated AI guidance through storytelling rather than commands Designed a reusable architecture suitable for games and XR
What we learned Less AI output increases immersion Constraints make AI feel more intelligent and human Storytelling can function as a user interface Real-time engines like Unreal are ideal for narrative AI systems
What's next for EchoTurn Deeper spatial awareness using in-engine context Persistent narrative memory across sessions Multi-character AI perspectives Creator tools for designing narrative echo styles Expansion into XR, training, and experiential learning use cases.
*NOTE: This project is a windows 11 executable. Please download the zip from the google drive. Extract it and run EchoTurn executable. On the main menu, you need to enter Gemini Api key, Eleven Labs api key.
Built With
- elevenlabs
- gemini
- unreal-engine
Log in or sign up for Devpost to join the conversation.