could not select device driver "" with capabilities: [[gpu]].

[LordKayBanks]

screenshot below:

[jim60105]

Do you have gpu support for your docker?
Please try again after completing this guide:
https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html

[LordKayBanks]

thanks. i got it working.

[LordKayBanks]

One more thing, How can I use the large v2 model (mult-language model), so that i dont have to run and download another docker image over and over whenever i need to transcribe for a different language?

On Server:

On collab:

As you can see, It did not re-download in this case

[jim60105]

On your collab, it did not re-download the align model because it had been downloaded before and the model is stored on the machine.
As for Docker, every time you excute docker run, it will open a new container and its cache is not shared.
You can docker mount the /cache folder to solve this problem.

[jim60105]

Oh, I think I should add this to the readme.

[LordKayBanks]

thank you. I should have thought of that.

persisted cache
docker run --gpus all -it -v ".:/app" -v whisper_cache:/cache ghcr.io/jim60105/whisperx:no_model -- --model large-v2 --language en --output_format srt audio.mp3