A machine learning project that converts songs into visualizations and extracts 6 different audio features per song. Features are used to create an algorithm to classify Spotify songs into ten different genres.
- An extensive walkthrough of our code/dependencies/contributions can be found here: https://audio-recognition.herokuapp.com/
- We recommend viewing all ipynb files in google collab due to cross platform librosa support!
- Be sure to upload "numerical_machine_data_FINAL.csv" to your google drive if you wish to run machine learning code
- Mount the drive to the correct directory within the "Deep_learning_model.ipynb" file before running
- WE DO NOT RECOMMEND RUNNING LIBROSA CODE! READ ONLY!!! (Unless you have 45 minutes or so to kill)
- Internet browser installed if you want to view the website or collab files!
Seth Abbott- Advanced librosa plotting, Audio Feature Extraction, Data Pre-Processing, Algorithm tuning/generation William Forsyth- Advanced Web Design, Advanced CSS, API call Caitlyn Calsbeek- API call, Mp3 pull, Advanced librosa plotting, Audio Feature Extraction Kathryn Panger- Advanced data clean-up for machine learning pre-processing, API call, Mp3 pull Danne Paredes- Web design, Heroku app deployment Heesung Shim- API call, Mp3 pull, Algorithm tuning/generation
- "Spotipy" (Spotify's API) used to pull song data and sample tracks
- Google Collab used for librosa, pre-processing, and machine learning
- Dependencies listed on herokuapp