2026-05-14 05:39 UTC
@latenightlinux@mastodon.social thanks for an interesting episode, I liked Will's take on Speech-to-text. I did notice he said he pressed scroll lock, talked, and then waited for a transcription. He does not need to do that: I've cobbled together a wayland client for live transcription using local (whisper) model, here's a link to the repo: https://github.com/bjodah/whisper_streaming (I must admit I've only used the gnome extension so far, so the non-gnome wayland client is untested...).
Replies (2)
-
@bjodah@mastodon.social 2026-05-14 05:41
I should probably add a video showing gnome intergration, in the mean time I uploaded one for the emacs-client (in the same repo), its mode of operation is very similar: https://www.reddit.com/r/emacs/comments/1sne3vf/local_live_transcription_in_emacs_speechtotext/
-
@8none1@fosstodon.org 2026-05-14 09:53
@bjodah@mastodon.social @latenightlinux@mastodon.social I’ll give that a go and talk about how I get on, thanks