search

R3gm / SoniTranslate

Synchronized Translation for Videos. Video dubbing
Apache License 2.0
806 stars 150 forks source link

Add Docker #109

Open ankelkoss opened 6 days ago

ankelkoss commented 6 days ago
  1. Install driver from README
  2. Create 2 files: Docker and entrypoint.sh
  3. Copy the data to each file respectively
  4. In the docker file, update the token variable: YOUR_HF_TOKEN And if need OPENAI_API_KEY (remove the "#" at the beginning of the line)
  5. If you have a problem with cuDN
  6. In powershell (with dot) > 
    docker build -t soni_translate_image .
docker run -it --gpus all -p 80:7860 --name sonitranslate soni_translate_image
  7. After successful compilation (~30 min) and launch, SoniTranslate will be available at http://127.0.0.1/
  8. You will need approximately: 8.1. ~12 gigabytes of RAM (at the compilation stage ~12-20), 8.2. ~30-50 gigabytes of free disk space, 8.3. Good video card, 8.4. Fast internet

Dokerfile

# Use a base image with CUDA support
FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04

# Install the necessary tools
RUN apt-get update && \
    apt-get install -y --no-install-recommends wget git ffmpeg && \
    rm -rf /var/lib/apt/lists/*

# Install Miniconda
RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda.sh && \
    bash miniconda.sh -b -p /opt/conda && \
    rm miniconda.sh && \
    /opt/conda/bin/conda init

# Add the path to conda to the environment
ENV PATH=/opt/conda/bin:$PATH

# Create and activate the conda environment
RUN /opt/conda/bin/conda create -n sonitr python=3.10 -y && \
    /opt/conda/bin/conda run -n sonitr pip install pip==23.1.2

# Specify the working directory
WORKDIR /app

# Clone the SoniTranslate repository
RUN git clone https://github.com/r3gm/SoniTranslate.git

# Install the necessary compilers and tools
RUN apt-get update && apt-get install -y build-essential cmake nano

# Install the requirements_base.txt dependencies
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -r requirements_base.txt -v

# Install the dependencies requirements_extra.txt
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -r requirements_extra.txt -v

# Install onnxruntime-gpu
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install onnxruntime-gpu

# Install dependencies requirements_xtts.txt
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -q -r requirements_xtts.txt

# Install TTS 0.21.1 without dependencies
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -q TTS==0.21.1 --no-deps

# Remove old versions of numpy, pandas and librosa
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip uninstall -y numpy pandas librosa

# Install the required versions of numpy, pandas and librosa
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install numpy==1.23.1 pandas==1.4.3 librosa==0.10.0

# Install the required versions of tts and torchcrepe
RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install "tts<0.21.0" "torchcrepe<0.0.20"

# Set up environment variables in conda
RUN /opt/conda/bin/conda run -n sonitr conda env config vars set YOUR_HF_TOKEN="INSERT_TOKEN_HERE"

# Set up environment variables in conda
#RUN /opt/conda/bin/conda run -n sonitr conda env config vars set OPENAI_API_KEY="INSERT_TOKEN_HERE"

# Modify app_rvc.py to add server_name="0.0.0.0" after max_threads=1
RUN sed -i '/app\.launch(/,/debug=/s/max_threads=1,/max_threads=1, server_name="0.0.0.0",/' /app/SoniTranslate/app_rvc.py

# Open port 7860 in container
EXPOSE 7860

# Copy entrypoint.sh to container
COPY entrypoint.sh /app/entrypoint.sh

# Go to repository directory
WORKDIR /app/SoniTranslate

# Command to run Python application via entrypoint.sh
CMD ["/bin/bash", "-c", "/app/entrypoint.sh"]

entrypoint.sh

#!/bin/bash

# Activate the 'sonitr' environment
source /opt/conda/etc/profile.d/conda.sh
conda activate sonitr

# Run the Python application
python /app/SoniTranslate/app_rvc.py
biraescudero commented 4 days ago

Thank you for your detailed instructions.

Em sáb., 26 de out. de 2024 10:02, ankelkoss @.***> escreveu:

  1. Install driver from README
  2. Create 2 files: Docker and entrypoint.sh
  3. Copy the data to each file respectively
  4. In the docker file, update the token variable: YOUR_HF_TOKEN And if need OPENAI_API_KEY (remove the "#" at the beginning of the line)
  5. If you have a problem with cuDN https://github.com/R3gm/SoniTranslate/issues/79#issuecomment-2375711377
  6. In powershell (with dot) >

docker build -t soni_translate_image . docker run -it --gpus all -p 80:7860 --name sonitranslate soni_translate_image

  1. After successful compilation (~30 min) and launch, SoniTranslate will be available at http://127.0.0.1/
  2. You will need approximately: 8.1. ~12 gigabytes of RAM (at the compilation stage ~12-20), 8.2. ~30-50 gigabytes of free disk space, 8.3. Good video card, 8.4. Fast internet

Dokerfile

Use a base image with CUDA support

FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04

Install the necessary tools

RUN apt-get update && \ apt-get install -y --no-install-recommends wget git ffmpeg && \ rm -rf /var/lib/apt/lists/*

Install Miniconda

RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda.sh && \ bash miniconda.sh -b -p /opt/conda && \ rm miniconda.sh && \ /opt/conda/bin/conda init

Add the path to conda to the environment

ENV PATH=/opt/conda/bin:$PATH

Create and activate the conda environment

RUN /opt/conda/bin/conda create -n sonitr python=3.10 -y && \ /opt/conda/bin/conda run -n sonitr pip install pip==23.1.2

Specify the working directory

WORKDIR /app

Clone the SoniTranslate repository

RUN git clone https://github.com/r3gm/SoniTranslate.git

Install the necessary compilers and tools

RUN apt-get update && apt-get install -y build-essential cmake nano

Install the requirements_base.txt dependencies

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -r requirements_base.txt -v

Install the dependencies requirements_extra.txt

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -r requirements_extra.txt -v

Install onnxruntime-gpu

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install onnxruntime-gpu

Install dependencies requirements_xtts.txt

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -q -r requirements_xtts.txt

Install TTS 0.21.1 without dependencies

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install -q TTS==0.21.1 --no-deps

Remove old versions of numpy, pandas and librosa

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip uninstall -y numpy pandas librosa

Install the required versions of numpy, pandas and librosa

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install numpy==1.23.1 pandas==1.4.3 librosa==0.10.0

Install the required versions of tts and torchcrepe

RUN cd /app/SoniTranslate && /opt/conda/bin/conda run -n sonitr pip install "tts<0.21.0" "torchcrepe<0.0.20"

Set up environment variables in conda

RUN /opt/conda/bin/conda run -n sonitr conda env config vars set YOUR_HF_TOKEN="INSERT_TOKEN_HERE"

Set up environment variables in conda

RUN /opt/conda/bin/conda run -n sonitr conda env config vars set OPENAI_API_KEY="INSERT_TOKEN_HERE"

Modify app_rvc.py to add server_name="0.0.0.0" after max_threads=1

RUN sed -i '/app.launch(/,/debug=/s/max_threads=1,/max_threads=1, server_name="0.0.0.0",/' /app/SoniTranslate/app_rvc.py

Open port 7860 in container

EXPOSE 7860

Copy entrypoint.sh to container

COPY entrypoint.sh /app/entrypoint.sh

Go to repository directory

WORKDIR /app/SoniTranslate

Command to run Python application via entrypoint.sh

CMD ["/bin/bash", "-c", "/app/entrypoint.sh"]

entrypoint.sh

!/bin/bash

Activate the 'sonitr' environment

source /opt/conda/etc/profile.d/conda.sh conda activate sonitr

Run the Python application

python /app/SoniTranslate/app_rvc.py

— Reply to this email directly, view it on GitHub https://github.com/R3gm/SoniTranslate/issues/109, or unsubscribe https://github.com/notifications/unsubscribe-auth/AL4Y2XDHJETRMHOO7R7ZEALZ5OHGTAVCNFSM6AAAAABQUZHNSCVHI2DSMVQWIX3LMV43ASLTON2WKOZSGYYTKOBVGM4DKNI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

gu-ma commented 21 hours ago

Thanks @ankelkoss, there's also an alternative using the hf space:

  1. Clone repo and cd into it
  2. Add server_name='0.0.0.0' to app.launch() in app_rvc.py
  3. Create a docker-compose.yml
  4. Run docker-compose up (add -d to run it detached)

docker-compose.yml:

services:

  sonitranslate:
    image: registry.hf.space/r3gm-sonitranslate-translate-audio-of-a-video-content:latest
    environment:
      - YOUR_HF_TOKEN="INSERT_TOKEN_HERE"
      - OPENAI_API_KEY="INSERT_TOKEN_HERE"
    working_dir: /app
    volumes:
      - .:/app
      - ~/.cache:/root/.cache
    ports:
      - 80:7860
    command: python app_rvc.py
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

More info on running hf spaces in Docker