The music category winner of the HackXelerator 2025 sprint was SoundMatch, a solo project developed by Birdmania. With AI-powered vibe matching and rich data aggregation, SoundMatch helps creators discover royalty-free music that fits the feel of their content without hours of manual searching or legal ambiguity.
A solo sprint born from a real-world pain point
Birdmania entered the HackXelerator with one mission: turn a side project idea into something tangible. The inspiration? A friend who constantly struggled to find good, royalty-free tracks that matched the vibe of their videos. After hearing one too many complaints about the painful search process, Birdmania saw an opportunity.
SoundMatch set out to solve a deceptively simple problem: How do you help creators find music that fits their intended mood quickly, affordably, and without copyright headaches?
A pragmatic AI stack built in 20 days
Birdmania used the 20-day sprint to build a fully functional MVP using Python and FastAPI. The backend integrated AcoustID for audio fingerprinting, Jamendo for royalty-free music sourcing, and Gemini Pro for AI-powered vibe and keyword matching.
You can search by artist or track name, upload an MP3, or drop in a YouTube link. SoundMatch then pulls metadata from sources like MusicBrainz and Discogs, uses Gemini to identify moods and themes, and matches those against royalty-free tracks with similar sonic qualities. The result is a clever shortcut for creators who want to find “the right feel” without spending hours scrolling through generic libraries.
Building this wasn’t exactly smooth sailing. Spotify’s API was, in Birdmania’s words, “just beyond trash” for their needs, and Shazam’s was inaccessible. The workaround? Fall back on AcoustID and leverage YouTube’s open ecosystem. AI prompt tuning also proved tricky: Gemini nailed Enya-style moods but stumbled hard on Travis Scott at first.
The MVP worked together despite the API chaos and some async pipeline headaches. That alone felt like a win.
What’s next for SoundMatch
Birdmania is far from done. Plans for future development include:
- Adding more royalty-free sources like Audius and Free Music Archive
- Implementing deeper local audio analysis using tools like librosa or Essentia
- Introducing vector search with FAISS for improved similarity matching
- Exploring integration with DAWs like Ableton Live to support creative workflows
- Refining Gemini prompt engineering for more accurate and consistent mood tagging
For now, SoundMatch is a passion project with clear potential to grow into something bigger.
Why the HackXelerator worked
The sprint format provided just enough structure and time pressure to force progress. Birdmania worked solo, didn’t lean on mentorship, and focused hard on shipping something functional. The result? An innovative tool that scratches a very real itch for creators, which took home the win for music in the process.