Aug 2019 – Jan 2021 · Atlanta, GA
Fynda
Deep Learning & Full Stack
Music taste mapping with ML genre classification from audio.
- Role
- Deep Learning & Full Stack
- When
- Aug 2019 – Jan 2021
- Where
- Atlanta, GA
- Focus
- AI
01 · The problem
Music discovery apps recommend by popularity, not by what a song actually sounds like.
02 · What I built
Classified genre from raw audio with TensorFlow/Keras, Kapre and Librosa, combined with Spotify APIs to build taste maps, genre ratios and event discovery in React.
03 · Outcome
A working taste-mapping experience driven by audio ML rather than charts.
Highlights
- ✦Audio-based genre classification
- ✦Spotify API taste maps
- ✦Event discovery
What I did
- ✓Genre classification from audio samples using Kapre and Librosa.
- ✓Taste maps, genre ratios and event discovery in React.
Toolbox
TensorFlowKerasLibrosaSpotify APIReact
~ waveforms ~