-
Notifications
You must be signed in to change notification settings - Fork 289
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
Chrome extension update #123
Conversation
Extension rewrite
Hello, thanks for putting this together, can you do a quick merge with main and we can start looking into this. |
So like this? Sorry, still learning stuff. |
whisper_live/server.py
Outdated
@@ -154,7 +155,7 @@ def recv_audio(self, | |||
options["model"] = faster_whisper_custom_model_path | |||
client = ServeClientFasterWhisper( | |||
websocket, | |||
multilingual=options["multilingual"], | |||
multilingual=False, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We dont need this as well. This can be removed.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We can run the FasterWhisper backend without multilingual, this can be removed from this backend. So we can remove these lines as well:
WhisperLive/whisper_live/server.py
Line 368 in 60455b1
multilingual=False, |
WhisperLive/whisper_live/server.py
Line 157 in 60455b1
multilingual=options["multilingual"], |
Which brings me to the client, and client already sends the language with the model so, it doesn't need to worry about sending the multilingual option as well. If you don't mind updating the client i.e. removing the multilingual option from there that would be really appreciated, else we can take a look at it as well.
The TensorRT backend would need it for the tokenizer.
Audio-Transcription-Chrome/popup.js
Outdated
@@ -129,7 +119,6 @@ document.addEventListener("DOMContentLoaded", function () { | |||
startButton.disabled = isCapturing; | |||
stopButton.disabled = !isCapturing; | |||
useServerCheckbox.disabled = isCapturing; | |||
useMultilingualCheckbox.disabled = isCapturing; | |||
modelSizeDropdown.disabled = isCapturing; | |||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
taskDropdown.disabled = isCapturing | |
I guess it was missed initially.
…per backend. Disable task dropdown when capturing in chrome extension.
I didn't even notice that drop-down wasn't getting disabled. I haven't looked into tokenizer but faster whisper is choosing it here and it gets multilingual based on the model right? That's why we don't care about passing it. |
self.client_uid
inrecv_audio
was being accessed during exception without being declaredServeClientFasterWhisper
self.multilingual
get_model_size
function to justcheck_valid_model
since we don't need the multilingual part anymore.I'm new to this so any advice is appreciated.
I've checked the extension on different combinations of languages and models and didn't find any problems.
The
multilingual
arg is still used in some places and I don't know what you've planned with it so I left it inserver.ServeClientFasterWhisper
.If extension changes get approved I'll do the Firefox one.