martin-opensky / whisper-assistant-vscode Goto Github PK

This occurs on Windows 10.

unicodeencodeerror: 'charmap' codec can't encode character '\u3002' in position 8681: character maps to

due to this line in the --help file

parser.add_argument("--append_punctuations", type=str, default=""'.。,，!！?？:：”)]}、", hel ...

I fixed mine by going to transcribe.py and removing those characters. I suggest you add some broader character encoding, or somehow ignoring the stdout and simply checking for errors.

I have a functioning local Whisper in a Conda ENV. How can i direct this plugin to use it?

There is no question as the whether this installation is working, as i use it daily with ObsidianMD.
I may simply be missing how to configure the plugin.

Thank you

Got the extension 'working', but recording seems to be trying to save where I am running my cursor projects out of. Also it is building the path with forward and backward slashes --

A system error occurred (ENOENT: no such file or directory, unlink 'c:...\Coding Tools\My Projects/.whisper/recording.wav')

SoX is in the path, and can be accessed. Plugin does not detect it. Please expose a path.

Thank you

I was testing this extension and found a weird error where I have installed SoX via ARM-based homebrew, but the whisper assistant isn't picking up on it.

Any tips on debugging? I've restarted cursor/vscode a couple times but to no avail.

Attached is a screenshot of SoX not being detected, despite the path and version being present on my system.

Hello, this looks really good but uploading voice data to the cloud is a big no for me at least.

I was thinking that maybe it'd be a good idea to offer the ability to use something like https://github.com/Vaibhavs10/insanely-fast-whisper so that it can remain 100% local :)

Recorded a video.
https://www.loom.com/share/4e6bb1a7ff7542ca9bbb21b08767758e

Happy to help if you need more info