This repository provides WhisperX in a Docker image.
The image installs Python 3.10 on Ubuntu 22.10 base images, as well as pre-requisites such as PyTorch.
WhisperX can be ran on CPU or GPU, using either of the following Docker images.
docker run -it --rm -v /path/to/audio_files:/app thomasvvugt/whisperx:cpu recording.mp3 --batch_size 4 --compute_type int8
Note: You require a hardware graphics card (GPU), but this is much more performant.
docker run -it --rm --gpus all -v /path/to/audio_files:/app thomasvvugt/whisperx:cuda118 recording.mp3 --batch_size 8 --diarize --hf_token YOUR_HUGGINGFACE_READ_TOKEN
You can either use the pre-built images from Docker Hub, or build the images yourself.
docker build -f Dockerfile.cpu -t whisperx:cpu
docker build -f Dockerfile.cuda118 -t whisperx:cuda118