Turbo-V3

[dudztroyer]

Openai released Whisper Turbo V3. Can we get on this worker?

SYSTRAN/faster-whisper#1025

[Nik-Kras]

Try using "deepdml/faster-whisper-large-v3-turbo-ct2" as a model name

[dudztroyer]

@Nik-Kras this implementation preloads the model weights on docker image from the original weights. Doesn't work only by setting the name though.

[asr-lord]

You could include the following code in:
https://github.com/runpod-workers/worker-faster_whisper/blob/main/builder/fetch_models.py

from huggingface_hub import snapshot_download

repo_id = "deepdml/faster-whisper-large-v3-turbo-ct2"
local_dir = "faster-whisper-large-v3-turbo-ct2"
snapshot_download(repo_id=repo_id, local_dir=local_dir, repo_type="model")

[yccheok]

It seems we'll need to wait until the upstream project SYSTRAN/faster-whisper#1030 is updated with Turbo and published on https://pypi.org/project/faster-whisper/. Only then will this feature become available.