pluja / whishper

Transcribe any audio to text, translate and edit subtitles 100% locally with a web UI. Powered by whisper models!

Home Page:https://whishper.net

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[Feature] Support Whisper large-v3

RayBB opened this issue · comments

Description of the feature you are requesting:
Support the latest version of whisper https://news.ycombinator.com/item?id=38166965

Describe the solution you'd like
Be able to select it like the other models

Describe alternatives you've considered
n/a

Additional context
n/a

So, faster-whisper - the implementation of Whisper, that Whishper is utilizing - is not capable of using large-v3 yet. So we need to wait for that to happen.

There is already an issue about it SYSTRAN/faster-whisper#544 but I'm not sure if faster-whisper will migrate to large-v3 since its brilliant developer Guillaume Klein started a job as Machine Learning Engineer at Apple last month.

Keep an eye on this PR: SYSTRAN/faster-whisper#548

When merged, I will update to support it.

Already faster-whisper supported large-v3 and CPU device with parameter --device=CPU.

so SYSTRAN/faster-whisper#578 got merged instead.
So v3 is going to be added to whishper soon then?

Added in 82b531f

Now available in v3.1.0