Feature and Similarity based playlisting

This repository contains code to run through a given directory of mp3 files, analyse features, present the data, and run an interactive web-app to build playlists based on features and similarity.

Installation:

Create a python 3.10 environment in conda, and run pip install -r requirements.txt. Code was tested with python-3.10. Essentia is known to be troublesome to install on linux and windows, sorry about that.

Use:

All commands are meant to be run from the project directory.

There are 3 files to run:

1. analysis.py: This script scans a given folder for mp3 files and creates a json file that stores the data. Make sure to specify the paths in the top of the file before running. To run, execute python src/analysis.py. Note that this process may take some time.

2. data_analysispy: This script analyses the audio_features.json file and creates graphs of relevant data of the library. This script also creates audio_features.pkl for use in the playlisting web-app. To run, execute python src/data_analysis.py.

3. myApp.py: This script contains the code needed to run the web-app. To run, execute streamlit run src/myApp.py.