- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
You must log in or # to comment.
Vibe-coded trash + outdated models.
Wouldn’t have expected anything better from a trash website.
Fair if whisper.cpp tiny/base feels dated. Default is still on-device whisper.cpp with Vulkan, larger models and VOSK are in Settings, and we are adding Parakeet. Code is AGPL on GitHub.
Also, I am a backend person first, so the old site taste was on me. I have been working on that. Would love you to hit vocalinux.com again and tell me what still sucks. :)
trash website.
How do you mean? This is such a harsh criticism, I’m curious what you’re basing it on?
My biggest concern is that it suggests to Yolo pipe the bash script from the website into your terminal. And the way it suggests doing it (saving it to /tmp first and then executing it from /tmp) feels very LLM-y to me.
Yeah, pipe-to-bash is a bad habit even when the script is ours. Prefer git clone / AppImage / read install.sh first; the /tmp step is so you can open it before it runs, not so you skip reading it.
I always find that suspicious, yes. Not to be AI though — sites have been suggesting this for decades — but I feel like it speaks to the credibility and experience of the people behind the site. Even if AI wrote it, why is the human publishing it.
Everyone should know you should not run someone else’s script unless you understand it yourself, or can at least determine that there’s no malicious stuff happening in it. This should be common knowledge. And don’t copy paste command lines into terminal unless you know what it does.
Did we forget this in the *nix community?
I found this program not long ago and it does well as long as you adjust your mics. Turns out my built in laptop mic was peaking and gave poor inputs to the program. Another bottleneck is ram, my laptop also doesn’t have the most(8gb), even so it is able to use the larger models at a slower speed. I recommend it for anyone needing a privacy focused dictation to text program.
Glad the mic gain fixed it. 8 GB is tight for the bigger models, so sticking to tiny/base or quantized whisper.cpp is the sane path there. I’m also adding support for Parakeet models soon. That works incredibly well on CPU only machines with low RAM.
I’ve been using it on both mint and arch for probably almost a year. Generally works great apart from a couple of small things. On mint however it did voice typing and on Arch it just goes to clipboard and needs to be pasted. I haven’t looked for awhile but I think Anthropic have something to do with the coding of it. Article is blocked so can’t read that.
We have pushed a lot of Ibus/Wayland/X11 fixes in last couple of months. Would you mind trying it again on Arch? It should be able to inject it properly now.
Sounds interesting to use together with a Note taking application.
Works fine next to a notes app. Dictation goes into whatever has focus.
