This is the code for the paper Composing Four-Part Bach Harmonies with Evolutionary Algorithms.
In order to view the generated score, you need to have MuseScore 4 installed on your machine. Please refer to https://musescore.org/en/handbook/3/installation. You also need Python 3.x+
- Clone the repo:
git clone https://github.com/stavaser/bach-chorale-harmonizer.git - Navigate to the directory:
cd bach-chorale-harmonizer - Create a virtual environment:
python3 -m venv venv - Active the env on macOS/Linux:
source venv/bin/activate - Install dependenices:
pip3 install music21
pip3 install pygame
pip3 install argparse
- Set the env variable in
utils.py:- ensure
MUSESCORE_PATHis the same as the MuseScore path on your machine
- ensure
Run
python3 main.py --verbose=True --population_size=50 --generations=10 --encourage_diversity=True --use_intervalic_qualities=True --use_musescore=True
xmlfile: Path to your XML file. Use this to override the random Bach selection. If not provided, a random Bach chorale is selected.verbose: Boolean flag for verbose output. Default is True.population_size: Integer specifying the population size for the evolutionary algorithm. Default is 50.generations: Integer specifying the number of generations for the evolutionary algorithm. Default is 10.encourage_diversity: Boolean flag to encourage diversity in the evolutionary process. Default is True.use_intervalic_qualities: Boolean flag to use intervalic qualities in the fitness evaluation. Default is True.use_musescore: Boolean flag to use MuseScore for output. If False, the script will play the audio directly. Default is True.
Run the harmonizer with default settings:
python3 main.py
Run the harmonizer with a specific XML file and custom settings:
python3 main.py --xmlfile=./path/to/file.xml --verbose=False --population_size=100 --generations=20 --encourage_diversity=False --use_intervalic_qualities=False --use_musescore=False
Adjust these weights to influence the style and characteristics of the generated chorale.
| Weight Key | Default Value | Description |
|---|---|---|
eval_valid_progression |
15.0 | Prioritizes the validity of chord progressions. Higher values emphasize adherence to traditional harmonic progression rules. |
eval_parallel_octaves |
1.0 | Penalizes parallel octaves. Increasing this value reduces the occurrence of parallel octaves in the harmonization. |
eval_parallel_fifths |
1.0 | Penalizes parallel fifths. A higher value discourages the use of parallel fifths. |
eval_parallel_unisons |
1.0 | Penalizes parallel unisons. Increasing this weight avoids parallel unisons in the harmonization. |
eval_tritones |
10.0 | Limits the usage of tritones, often considered dissonant. A higher weight will reduce tritones in the harmonization. |
eval_repeated_bass |
1.0 | Addresses the repetition of the bass note. Higher values discourage excessive repetition of the same bass note. |
eval_voice_ranges |
1.0 | Ensures that each voice stays within its comfortable singing range. |
eval_voice_closeness |
1.0 | Evaluates the closeness of voices. Promotes more spacing between the voices with higher values. |
eval_voice_crossing |
1.0 | Penalizes instances where voices cross each other. Higher values discourage voice crossing. |
eval_intervalic_quality |
1.0 | Examines the intervalic quality between the Soprano line and a generated Bass line. The score is determined based on a probability matrix created from analyzing hundreds of Bach chorales, evaluating every possible interval between Soprano and Bass lines. |
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent