Built double-tap-Ctrl dictation for Linux on nemotron-speech-streaming-en-0.6b #16116
NelsonScott
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
I built a macOS style dictation tool for Fedora/GNOME Wayland on top of nemotron-speech-streaming-en-0.6b and the cache aware streaming API, and it came out great. Double tap Ctrl, talk,and words type into whatever window has focus about half a second behind your voice, all local.
Its emitting real keystrokes so I can never revise a word after typing it. Running 160ms hops with att_context [70,1] via conformer_stream_step, about 1.5 GB VRAM.
Code (MIT): https://github.com/NelsonScott/speakwrite
Was actually my first time using this mode and it worked out great for this project, thanks for shipping a streaming model at this size!
All reactions