SoniTranslate: video translation web application

Getting started

1. 1

   The SoniTranslate README's Before You Start section lists installing the NVIDIA drivers for CUDA 11.8.0 and accepting the license agreement for using Pyannote.

2. 2

   A token must be created to use some of the automatic model download features in SoniTranslate, and in Linux the Hugging Face token can be set as an environment variable.

3. 3

   The installation steps create and activate an anaconda environment for SoniTranslate, clone the github repository and navigate to it, and install the required packages.

4. 4

   Installing ffmpeg with Anaconda is recommended, it is needed to process audio and video files, and optional packages can be installed after installing FFmpeg.

5. 5

   To run locally, make sure the sonitr conda environment is active, and when the local URL is displayed in the terminal, open it in a web browser to access the SoniTranslate interface.

6. 6

   In most environments, pressing Ctrl+C in the terminal where SoniTranslate was launched stops the execution, and a command deactivates the Conda environment.

conda create -n sonitr python=3.10 -y
conda activate sonitr
python -m pip install pip==23.1.2 Setuptools==80.6.0
conda install pytorch==2.5.1 torchvision==0.20.1 torchaudio==2.5.1 pytorch-cuda=11.8 -c pytorch -c nvidia

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

pip install -r requirements_base.txt -v
pip install -r requirements_extra.txt -v
pip install onnxruntime-gpu

conda activate sonitr

python app_rvc.py

conda deactivate