Inspiration
This project was inspired by a love for global cultures and the way people speak in different parts of the world. We wanted to go beyond boring, generic AI chats by building an experience that felt real, fun, and deeply connected to the way people actually talk. It started with two characters—"Toronto Yute" and "Jamaican Uncle"—and grew to include a “UK Roadman.” The goal? To build a chaotic, hilarious, and authentic voice chat playground where users could hear these cultural voices come to life.
We built the app using React with TypeScript for structure and logic, Tailwind CSS for fast and clean styling, and JavaScript/HTML to power the frontend. For voice, we integrated ElevenLabs for realistic voice cloning and combined it with Gemini AI to generate dynamic, in-character responses. The mix of custom prompts and real-time voice synthesis created an engaging, human-like chat experience.
What it does
We built Cultural Voice Chat AI to explore how language, slang, and personality can come alive through tech. Inspired by the unique speech styles of different cultures, we wanted to create an immersive chat experience where users talk to AI characters that sound and act real.
Using React, TypeScript, Tailwind CSS, and JavaScript, we built the frontend. For voice, we integrated ElevenLabs for cloning realistic accents, and Gemini AI for generating character-specific, slang-rich responses. The result is a chaotic, hilarious, and authentic voice playground.
What we learned
How to blend AI text and voice seamlessly
The importance of detailed prompt engineering
Managing real-time state and voice playback smoothly
Tuning AI safety filters to preserve cultural authenticity
Log in or sign up for Devpost to join the conversation.