vibefm

We extract the formula behind your track, you discover its acoustic DNA.

Comment

Inspiration

  • Shazam and statsfm are really cool apps for music listeners, Shazam allows you to figure out songs using your microphone while statsfm(previously spotistats) allow you to track your Spotify listening activity and also give your song recommendations based on your listening data also having a social feature to allow to connect with people with similar taste profiles.

What it does

  • The aforementioned apps are amazing at what they do, but for accurate recommendations you need to rigorously analyze the songs' DNA.
  • This is VibeFM's differentiator, we extract your tastes and present them to you to help you find more of what you like and to connect with people who share your "Acoustic DNA".
  • This "Acoustic DNA" allows producers to better understand what makes something viral when analyzing the charting songs' data while also allowing people to meet people with similar tastes through our vector based matchmaking.

How we built it

  • Python and TypeScript
  • Expo SDK 54 for universal applications using React Native
  • RapidAPI(Shazam)
  • FreqBlog API
  • Gemini Flash(via OpenRouter)
  • Apple iTunes Search API for reliable music metadata
  • @siteed/expo-audio-stream for low-latency audio capture(for Shazam)
  • GPS for detecting local
  • BeautifulSoup(scrape Kworb.net for Global and local)

Challenges we ran into

  • Expo SDK version incompatibilities(non functional work)
  • Can't upload audio files to Shazam(API strictly expects raw PCM audio bytes, no built-in MP3, M4A, general audio decoders)
  • Tapic Engine disabled by default when microphone is active -> Wait for the mic to unlock the engine and send the pulses after recording

Accomplishments that we're proud of

  • First hackathon using React Native and Expo(especially for mobile apps)
  • Full-Stack Development experience

What we learned

  • React Native and Expo
  • RapidAPI
  • PCM(Pulse-Code Modulation, audio recording technology)

What's next for vibefm

  • Using technology like Higgsfield's "Virality Predictor" or Meta's "TRIBE V2" for brain simulations for even more complete "Acoustic DNA", matching brain heatmaps to moods and vibes.
  • Use more sources and APIs for a richer dataset for both trend prediction
  • Mapping sound frequencies to instruments, moods and vibes
  • Bigger toolkit for B2B producers/creators/labels; dashboards, complete description of extracted song DNA for actionable data.
Share this project:

Updates