Our server:
⚠️ Our server (for now) is too weak for Neural Generation. LSTM models take up to 5 minutes (instead of 30 seconds) per generation. GPT2 models aren't available. Run the server locally to test these models.
The instructions below are for Linux only and have been tested on Ubuntu 20.04 or later.
- Build a Docker Image:
sudo docker build -t deploy-reso .
- Run a Docker container:
sudo docker run -p 8000:8000 deploy-reso
- Upgrade pip and install Python dependencies:
pip install --no-cache-dir --upgrade -r requirements.txt
- Install system-wide packages:
apt-get install -y --fix-missing fluidsynth ffmpeg musescore3 libqt5core5a timidity
- Run a local server:
uvicorn main:app --host 127.0.0.1 --port 8000
- Click Start generating for free.
- Choose a music engine (algorithmic / neural).
- Specify the desired properties of the composion.
- Listen to this!
For local testing (you don't need to run the server beforehand):
python3 run_tests.py --port 8000 --run-server
For more info run python3 run_tests.py --help
Note that we use Selenium, which requires Google Chrome to be installed on your machine. You can install it with the following commands:
wget https://dl.google.com/linux/direct/google-chrome-stable_current_amd64.deb
dpkg -i google-chrome-stable_current_amd64.deb; apt-get -fy install
No cherry-picking
- Model: Waltz. Tempo: Fast. Scale: C.
10577062.mp4
- Model: Etude. Tempo: Normal. Scale: E.
37639085.mp4
- Model: Calm Music. Tempo: Slow. Scale: F#.
73491010.mp4
- Model: GPT2-Chopin. Tempo: Normal. Scale correction: no.
57415558.mp4
- Model: LSTM-Mozart. Tempo: Fast. Scale correction: no.