Inspiration
In multicultural spaces like the U.S., social events bring together people from vastly different cultural backgrounds. But music — a central part of any gathering — often defaults to Western-centric or mainstream genres, leaving some guests disconnected. What if AI could not only recognize but represent everyone in the room — through music? That led to Represent, an intelligent, culturally inclusive DJ that makes every party feel like home for everyone.
What it does
Represent curates a seamless song mix that reflects the diverse cultural backgrounds of the people at your event. It:
- Analyzes the cultures present at a party
- Curates songs that authentically represent each culture
- Arranges them for optimal musical flow
- Analyzes transitions between songs like a pro DJ
- Adapts the energy curve based on event type (e.g., warm-up → peak → cool-down)
- Outputs a professional, blended mix with transition timestamps — or a Spotify playlist
Every guest feels represented. Every track has a place. Every vibe is intentional.
How we built it
Built a multi-agent AI system, with each agent mimicking a role in a professional DJ team: Cultural Expert Maps each user-input culture to authentic genres using Spotify’s genre metadata. Song Curator Searches for culturally relevant songs via Spotify APIs, then arranges them for balance and flow. Transition DJ Analyzes adjacent track pairs using tempo, key, and energy to decide transition timing and method. Mood Analyst Designs the set's emotional curve based on energy level and party type (e.g., wedding, dinner, party).
Other tools:
- Spotify API for track search and audio feature analysis
- Pydub for audio stitching and transition blending
- Streamlit for the frontend UI
- Python to orchestrate agents and data flow ## Challenges we ran into
- Started building with crew.ai and langchain tools along with openai. But ran into multiple dependency issues. So, decided to build custom agents and tools.
- Spotify doesn’t allow audio mixing — we had to find workarounds using previews or user-owned files.
- Designing transitions that feel natural across very different cultural genres required experimentation with tempos and key signatures.
Accomplishments that we're proud of
- Designed and implemented a working multi-agent orchestration system from scratch
- Created a smooth DJ-style transition engine using only metadata and open-source tools
- Made inclusivity more than a talking point — turned it into a core technical design goal
- Built a working demo that lets people feel seen and celebrated through music at their events!
What we learned
- Building inclusive AI is a systems-level design challenge, not just a data problem
- Multi-agent architectures allow modular reasoning and explainability
- Musical transitions are not trivial — professional-sounding mixes require both technical precision and creative rules
- Culture-aware design means respecting authenticity while managing dynamic preferences
What's next for Represent
- Add a vision-based mood detection agent that uses a webcam to analyze real-time party energy and adjust music live
- Incorporate ML models to predict the best transitions and optimize song ordering dynamically
- Improve language-to-genre mapping for underrepresented communities using NLP + global music metadata
- Build a mobile-first web app with collaborative guest check-ins to shape the mix in real time
- Explore licensing with copyright-safe tracks to allow full mix playback directly in the app
Log in or sign up for Devpost to join the conversation.