🎶 Generate automatic playlists that follow your tracks' mood and groove

You will need to build bpm-tools, libkeyfinder and keyfinder-cli.

Here is a list of the dependencies that are required to build Autotracks and its dependencies:

• build-essential
• cmake
• git
• libfftw3-dev
• libavutil-dev
• libavcodec-dev
• libavformat-dev

Runtime dependencies are as follows:

• ffmpeg
• flac
• id3v2
• sox
• vorbis-tools

A very naive Dockerfile is provided within the repository. It will produce a whopping ~850 MB image -- which clearly leaves room for improvement :-)

1. Build the image:

git clone https://github.com/khannurien/autotracks.git
cd autotracks
sudo docker build -t autotracks .

2. Run the image:

sudo docker run -v /path/to/audio/files:/tracks -v /path/to/save/playlist:/output autotracks

If Autotracks runs successfully, this will:

• recursively look for audio files under /path/to/audio/files and create a corresponding .meta file alongside each audio file if it doesn't exist yet;
• recursively look for .meta files under /path/to/audio/files and parse tracks metadata from them;
• create a playlist.m3u file under the /path/to/save/playlist directory.

Environment variables can be used to set custom paths to keyfinder-cli and bpm-tag if necessary. Copy example.env to .env and edit it to reflect your system's configuration, or set and export the variables from your shell before launching Autotracks.

Use pipenv to enter a correct virtual environment (read Pipfile for Python dependencies):

git clone https://github.com/khannurien/autotracks.git
cd autotracks
pipenv install

Call Autotracks with a playlist name and a directory or a list of files:

pipenv run python3 -m src.autotracks "my_playlist.m3u" tracks/

This will create a my_playlist.m3u file in the current directory.

If some tracks remain unused or generate errors, their names will be displayed after playlist generation. You can then append them manually to the playlist if you wish.

Note: Autotracks creates a .meta file alongside each track of the list. These files contain the track key and BPM and are not removed after generation, in order to keep audio analysis results cached for further work. They can be safely removed should you not need them anymore.

Run tests:

pipenv install --dev
pipenv run python3 -m pytest