Werewolf : Undercover

The village falls asleep..
Werewolf
Seer
Cupid
Little girl
Villager
Witch

The app

I’ve always loved social deduction games like Werewolf and Mafia, but setting them up with cards and explaining the rules was always too much friction. I wanted to build a reinterpretation that’s self-explaining, child-friendly, multilingual, and that truly creates an immersive ambiance.

Werewolf: Undercover is my modern take on these classics, reimagined as a single app:

The app automatically assigns roles.
A game master voice (powered by ElevenLabs’ immersive narration) guides everything — everyone plays, no rules to read, and no one has to sit out as the moderator.
Supports multiple languages so anyone can join, including children.
Original chilling music and immersive sound effects (via ElevenLabs) bring the village to life.

I built the iOS app in SwiftUI, python script for ElevenLabs API. I used the ElevenLabs API to automatically generate and import localized audio narration. For the ambiance, I also used ElevenLabs’ sound generator to create effects like werewolf growls, combined with original music to set the mood.

Challenges I ran into

Creating the complex algorithm that adapts roles and logic depending on the number of players.
Making the experience self-explanatory without instructions.
Handling voice synchronization across multiple languages.

Accomplishments that I’m proud of

Building a voice-driven game master with ElevenLabs that guides players from start to finish.
Automating multilingual narration and sound design through the ElevenLabs API.
Designing a reinterpretation of Werewolf/Mafia that connects people in real life, through debate and bluff.

What I learned

Voice narration is a powerful way to make a game self-explaining.
Simplifying the interface for children actually improved the UX for everyone.
Integrating scripted APIs for localization and sound effects can save huge amounts of manual work.

What’s next for Werewolf: Undercover