The LilypondToBandVideoConverter is a python script that orchestrates existing command line tools to convert a music piece written in the lilypond notation to
-
a PDF score of the whole piece,
-
several PDF voice extracts,
-
a MIDI file with all voices (with some preprocessing applied for humanization),
-
audio mix files with several subsets of voices (specified by configuration), and
-
video files for several output devices visualizing the score notation pages and having the mixes as mutually selectable audio tracks as backing tracks.
For processing a piece one must have
-
a lilypond fragment file with the score information containing specific lilypond identifiers, and
-
a configuration file giving details like the voices occuring in the piece, their associated midi instrument, target audio volume, list of mutable voices for the audio tracks etc.
Based on those files the python script -- together with some open-source command-line software like ffmpeg or fluidsynth -- produces all the target files either incrementally or altogether.
The tool-chain has several processing phases that can be run as required and produce the several outputs incrementally. The following figure shows the phases and their results and how the phases depend on each other.
The files (in yellow) are generated by the phases (in magenta), the configuration file (in green) and the lilypond fragment file (in blue) are the only manual inputs into the processing chain.
Those phases are:
-
extract: generates PDF notation files for single voices as extracts (might use compacted versions if specified),
-
score: generates a single PDF file containing all voices as a score,
-
midi: generates a MIDI file containing all voices with specified instruments, pan positions and volumes,
-
silentvideo: generates (intermediate) silent videos containing the score pages for several output video file kinds (with configurable resolution and size),
-
rawaudio: generates unprocessed (intermediate) audio files for all the instrument voices from the midi tracks,
-
refinedaudio: generates (intermediate) audio files for all the instrument voices with additional audio processing applied,
-
mix: generates final compressed audio files with submixes of all instrument voices based on the refined audio files with a specified volume balance and some subsequent mastering audio processing (where the submix variants are configurable), and
-
finalvideo: generates a final video file with all submixes as selectable audio tracks and with a measure indication as subtitle
The script and its components are written in python and can be installed as a single python package. The package requires either Python 2.7 or Python 3.3 or later.
Additionally the following software has to be available:
-
lilypond: for generating the score pdf, voice extract pdfs, the raw midi file and the score images used in the video files,
-
ffmpeg: for video generation and video postprocessing,
-
fluidsynth: for generation of voice audio files from a midi file plus some soundfont (e.g. FluidR3_GM.sf3),
-
sox: for instrument-specific postprocessing of audio files for the target mix files as well as the mixdown, and
Optionally the following software is also used:
-
qaac: the AAC-encoder for the final audio mix file compression.
-
mp4box: the MP4 container packaging software
The location of all those commands as well as a few other settings has to be defined in a global configuration file for the LilypondToBandVideoConverter.
Installation is done from the PyPi repository via
pip install lilypondToBandVideoConverter
Make sure that the scripts directory of python is in the path for executables on your platform.
A longer description is available here and the detailed manual is available here.