Python scripts for creating a soundtrack. Can also be used to create a soundtrack from audio files ripped from a game.
Despite the fact that this repo is called python-soundtrack-creator, the actual module is called soundtrack.
This will only work on python versions >=3.8 <3.12 (numba doesn't support 3.12 yet).
Clone the repository
git clone https://github.com/ego-lay-atman-bay/python-soundtrack-creator.git
cd python-soundtrack-creatorInstall the dependencies
pip install -r requirements.txtYou will also need ffmpeg, which can be installed with your favorite package manager.
If you are using Anaconda, install ffmpeg by calling
conda install -c conda-forge ffmpeg
If you are not using Anaconda, here are some common commands for different operating systems:
apt-get install ffmpeg
or
apt-get install gstreamer1.0-plugins-base gstreamer1.0-plugins-ugly
yum install ffmpeg
brew install ffmpeg
download ffmpeg binaries from this gyan.dev
You can tag all the audio files in a folder for tagging a soundtrack.
cd src
python -m soundtrack tag -husage: python -m soundtrack tag [-h] input
positional arguments:
input input soundtrack folder
options:
-h, --help show this help message and exit
--output OUTPUT, -o Output folder to put the soundtrack in. If ommitted, it will modify
the original files.
--artist ARTIST, -ar Artist or composer.
--title TITLE, -ti Track title. This can be regex to get the title from the filename.
--track TRACK, -tr Track number. This can be regex to get the track from the filename.
--band BAND, -b Album artist / band
--album ALBUM, -al Album title
--publisher PUBLISHER, -p
Publisher
--genre [GENRE ...], -g
Genre(s)
--disc DISC, -d Disc number. This can be regex to get the disc from the filename.
--cover COVER, -c Cover art image.
--clear Clear metadata before writing
--spreadsheet SPREADSHEET, -s
Add a metadata spreadsheet. This csv will have a header, and it will
match the first column then add the specified metadata tags.
All regex used grabs group 0, the default group (outside parenthesis). The regex will search the filename without the file extension.
Example:
python -m soundtrack tag "folder" --clear --title "(?![0-9]+)(?<= - ).*" --track "[0-9]+(?= - )" --album "album" --band "Album Artist" --publisher "Publisher" --cover "artwork.jpg"This will tag the following
filename,title,track,album,band,publisher
01 - title 1.mp3,title 1,1,album,Album Artist,Publisher
2 - title 2.flac,title 2,2,album,Album Artist,Publisher
15 - title 15.ogg,title 15,15,album,Album Artist,Publisher
The spreadsheet is a csv file, formatted with the , separator, and each line is a different track. Values can be surrounded by spaces, the script trims spaces off before using the values.
The first column is used to check what file to put the rest of the metadata on.
filename , artist , disc
track 1.mp3, artist 1, 1
track 2 , artist 2, 2
The first row matches the file track 1.mp3 and adds the tags {'artist':'artist 1', 'disc':'1'}
The second row matches the file track 2.wav and adds the tags {'artist':'artist 2', 'disc':'2'}
The first column matching is case insensitive, and if the column is filename, the extension may be omitted to match files of the same name, but different formats.
You can also set the first column to be a tag, so you can match by title instead.
title , composer , track
title 1, composer 1, 1
title 2, composer 2, 2
The first row matches a track with the title title 1 and adds the metadata {'composer':'composer 1', 'track': 1}
The metadata in the spreadsheet is set after all the other metadata, so track titles can be grabbed from the filename using regex, then metadata can be added from the spreadsheet using the title for the match.
You can also create a soundtrack from audio files ripped from a game, specifically to automate the looping process (and make it consistent). This is a powerful tool, but it does come with one drawback, you need to spend time setting up configuration files.
The main command to run is this
python -m soundtrack create "config.json"config.json includes all the information needed to create the soundtrack. Here is the format it uses (in json5 syntax). Any duplicate keys here just mean that they are different types of values you can put in.
{
"silence": { // add silence to start and / or end of track
"start": 0.1,
"end": 0,
},
"loop": { // customize the loop
"count": 2, // how many times to loop
"fade": { // customize the final fade
"function": "linear", // function to use. Currently only "linear".
"duration": 3, // Duration of fade.
"options": {
"start": 100, // volume to start the fade at
"end": 0, // volume to end the fade at
"fade-adjust": -100, // a value that modifies the curve
}
} // The fade is similar to the adjustable fade in Audacity. Currently there is not S-Curve. If you want to quickly test it out then you can either write some code, or open Audacity.
},
"tracks": { // required
"files": "path/to/folder/", // path to folder with the tracks
"files": { // track json to specify all the tracks here
"title": {
"track": "title.wav",
"tags": { // track metadata
"artist": "Artist"
}
}
},
"filename": "track.json" // track json filename to search for in the "files" folder.
},
"metadata": "metadata.csv", // metadata spreadsheet. Uses the same format as the `tag` command, the only difference is that there is no "filename" column, use "title" instead.
"metadata": {
"sheet": "metadata.csv", // metadata spreadsheet. Uses the same format as the `tag` command, the only difference is that there is no "filename" column, use "title" instead.
"tracks": { // track specific metadata (can also be specified in the track json)
"title": {
"track": 1
}
},
"tags": { // global metadata tags that will be filled in for every track
"album": "Album name",
"cover": "artwork.jpg",
"genres": ["Genre 1", "Genre 2"],
}
},
"output": "path/to/folder/", // required: Output directory to place the soundtrack in
"output": "path/to/folder/Disc {disc}/{track:02} - {title}.{format}", // output can also be a template that will be filled out with metadata for the track and file format.
"format": "flac", // file format to export to. TODO: add file format ffmpeg options
}It loops the number of times specified in "count", then loops it again, but the last loop will be trimmed to the duration, and then the fade will be applied. It will look something like this with 2 loops.
[intro][loop][loop][trimmed_faded_loop]
The track json defines all the tracks in the soundtrack. These can either be defined in the main config.json file, or in subfolders inside the folder specified in the main config.json.
An example with external track.json files.
{
"tracks": {
"files": "folder/",
"filename": "track.json"
}
}with the folder structure like this
- folder/
- track 1/
- intro.wav
- loop.wav
- track.json
- track 2/
- track.wav
- track.json
Alternatively, the info in the track.json files can be in the main config.json file.
{
"tracks": {
"files": {
"track 1": {
"track": {
"intro": "track 1/intro.wav",
"loop": "track 1/loop.wav",
}
},
"track 2": {
"track": "track 2/track.wav",
}
}
}
}With the file structure like this
- track 1/
- intro.wav
- loop.wav
- track 2/
- track.wav
Each track needs some config to configure how the track should loop (whether it should loop), and it's metadata. Files specified in the track.json config files are also relative.
{
"track": {
"track": "track.wav", // this will not loop. It can be used to use special editing, or on a track that doesn't loop.
},
"track 2": {
"track": {
"file": "track.wav",
}
},
"intro + loop": {
"track": {
"intro": "intro.wav",
"loop": "loop.wav",
}
},
"loop": {
"track": {
"loop": "loop.wav",
}
},
"single track loop": {
"track": {
"file": "track.wav",
"loop": "05165465", // starting sample
}
},
"overlay audio track": {
"track": [{ // use a list of dictionaries to overlay tracks, like adding the drums into the track.
"intro": "intro.wav",
"loop": "loop.wav",
},
{
"file": "drums.wav",
"loop": "123456",
}]
},
"track with metadata": { // by default this will be the track title
"track": "track.wav",
"tags": {
"title": "track title", // overrides the track title
"track": 5,
"artist": "artist",
}
}
}Any amount of tracks can be put into a single track.json file, but all of the track.json files inside the "files" folder will be used.
file structure
- config.json
- metadata.csv
- artwork.jpg
- tracks/
- track 1/
- intro.wav
- loop.wav
- track.json
- track 2/
- track.wav
- track.json
- track 3/
- intro.wav
- loop.wav
- intro (drums).wav
- loop (drums).wav
- track.json
- output/
config.json
{
"silence": {
"start": 0.1,
"end": 0,
},
"loop": {
"count": 2,
"fade": {
"function": "linear",
"duration": 8,
"options": {
"start": 100,
"end": 0,
"fade-adjust": -100,
}
}
},
"tracks": {
"files": "tracks/",
"filename": "track.json"
},
"metadata": {
"sheet": "metadata.csv",
"tags": {
"album": "Album name",
"cover": "artwork.jpg",
"genres": ["Video Games"],
}
},
"output": "output/{track:02} - {title}.{format}",
"format": "flac",
}track 1/track.json
{
"track 1": {
"track": {
"intro": "intro.wav",
"loop": "loop.wav",
}
}
}track 2/track.json
{
"track 2": {
"track": {
"file": "track.wav",
}
}
}track 3/track.json
{
"track 3": {
"track": [{
"intro": "intro.wav",
"loop": "loop.wav",
},
{
"intro": "intro (drums).wav",
"loop": "loop (drums).wav",
}]
}
}metadata.csv
title ,track, artist
track 1, 1, artist
track 2, 2, artist
track 3, 3, artist 2
after running
python -m soundtrack create "config.json"The new file structure will be
...
- output/
- 01 - track 1.flac
- 02 - track 2.flac
- 03 - track 3.flac