-
-
Notifications
You must be signed in to change notification settings - Fork 5.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Whisper Fix, js replacement of the gradio element to prevent browser crash. #6189
base: dev
Are you sure you want to change the base?
Conversation
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch (oobabooga#5257)
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
Merge dev branch
A js implementation of voice recording. The microphone gradio element was causing different errors in different browsers, causing them to eventually crash. oobabooga#5929 (comment) This code and the accompanying script.js files were originally gotten from: oobabooga@8ad2b65#diff-e2fbfb497940c19efe0964ba8d6656b6edad2ae579e155c8e294ec7850c76276 And edited by multiple LLMs until I got something working well.
A js implementation of voice recording. The microphone gradio element was causing different errors in different browsers, causing them to eventually crash. oobabooga#5929 (comment) This code and the accompanying script.py files were originally gotten from: oobabooga/text-generation-webui@8ad2b65#diff-e2fbfb497940c19efe0964ba8d6656b6edad2ae579e155c8e294ec7850c76276 And edited by multiple LLMs until I got something working well.
FIX for SuperboogaV2 when trying to unload databases "Clear Data" not working crashing textgen There is an issue with textgen when trying to unload a superboogav2 database. Creating a database works, but unloading does not. Issue created here: oobabooga#6181
This is necessary to get the updated libraries to work. I've tested this on Ubuntu, but not windows.
I've tested this in Ubuntu, but not windows
I've tested this in Ubuntu, but not MacOS
Blarp, I tried also making a pr for the superboogav2 database clear data fix. But I guess I can only do one at a time? Sorry I'm green at this. The fork also has updated superboogav2 code I tried doing as a different pull request, and accidently sent it to main instead of dev #6193 There is an issue with textgen when trying to unload a superboogav2 database. Creating a database works, but unloading does not. The database now unloads without error, although the number database embedding deleted is not properly reported out in the terminal. This is a fix to the issue here: #6181 |
It would be better to split the superboogav2 fixes into a separate PR, yes. That would make it easier to review the changes. |
I tried to do that originally, but when I tried pushing the second one to dev (after mistakenly pushing it to main), there was no option to do so, it kept pointing me back to my fork ¯_(ツ)_/¯ I can redo everything if you need me to, the whisper changes are these: The superboogav2 changes are these: |
Checklist:
This is my submission to fix Whisper_STT (a much loved extension <3 )
This is in reaction to this thread:
#5929
I've tested it with multiple devices running through different systems:
-android phone microphone through firefox, opera, and chrome
-desktop condenser microphone, through firefox, opera, and chrome
I think the STT models interrupt the sound more clearly as well with this code. I can now use the small english model where before I had to use the medium sized model to get accurate transcriptions.