Skip to content

Instantly share code, notes, and snippets.

@laurentbenaroya
Created December 27, 2023 21:44
Show Gist options
  • Save laurentbenaroya/af7d2378f6f211b9b75deae0f844cb1e to your computer and use it in GitHub Desktop.
Save laurentbenaroya/af7d2378f6f211b9b75deae0f844cb1e to your computer and use it in GitHub Desktop.
docker file for AICoverGen
# Utiliser l'image de base
FROM alex2612/fairseq_audioaug_150823:latest
# Mettre à jour le système et installer les paquets nécessaires
ENV DEBIAN_FRONTEND noninteractive
RUN apt-get update && apt-get install -y \
ffmpeg \
sox \
python3-scipy
# Copier requirements.txt et installer les dépendances Python
WORKDIR /app
COPY requirements_docker.txt .
RUN pip install --upgrade pip && pip install -r requirements_docker.txt
# Copier le répertoire local dans le conteneur
COPY . /app
# Commande par défaut à exécuter lors du démarrage du conteneur
CMD ["/bin/bash"]
Here are some files and explanations on how to use command-line code with docker.
In short, there's a Dockerfile based on a fairseq image.
I've modified the requirements.txt file a little.
It turns out that I installed something because Nvidia gpu was not supported in the container,
but in fact the `--gpus all` option should suffice.
Here the explanation :
1) download the base docker container
> docker pull alex2612/fairseq_audioaug_150823:latest
2) goto top directory in AICoverGen
in src/main.py, replace action=argparse.BooleanOptionalAction by action='action_store'
it is a python version compatibility problem.
3) create a folder XXX in rvc_models (replace XXX by whatever you want).
Copy your rvc model in rvc_models/XXX, and unzip it there (if needed)
4) (?) Install the NVIDIA Container Toolkit (IM NOT SURE THAT THIS IS NEEDED, PLEASE TRY FIRST WITHOUT THIS STEP)
# https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
5) build docker image based on Dockerfile
> docker build -t aicovergen .
6) run it !
> docker run --gpus all -it --name aicovergencontainer aicovergen
then inside the docker :
> python src/download_models.py
> python src/main.py -i my_youtube_url -dir XXX -p 0
7) finally in another terminal, copy the output files :
> docker cp yyy:/app/song_output mysong
where yyy is the container ID or name (`docker ps` will give you yyy)
deemix
faiss-cpu==1.7.3
ffmpeg-python>=0.2.0
gradio==3.39.0
lib==4.0.0
librosa==0.9.1
numpy==1.23.5
onnxruntime_gpu
praat-parselmouth>=0.4.2
pedalboard==0.7.7
pydub==0.25.1
pyworld==0.3.4
Requests==2.31.0
soundfile==0.12.1
torchcrepe==0.0.20
tqdm==4.65.0
yt_dlp==2023.7.6
sox==1.4.1
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment