dictate : Local Offline Voice-to-Text for Claude Code on Linux #981
Description
Display Name
dictate
Category
Tooling
Sub-Category
General
Primary Link
https://github.com/vimalk78/dictate
Author Name
Vimal Kumar
Author Link
License
MIT
Other License
No response
Description
Local offline voice-to-text for Claude Code on Linux using faster-whisper. Daemon keeps the Whisper model loaded for instant transcription — no startup delay, no cloud APIs, no API keys. Includes a /dictate slash command, a voice-enabled editor with F5/F6 record keys, per-project vocabulary hints for technical terms, and push-to-talk with clipboard output. Can offload transcription to a remote GPU over LAN. Tested on Fedora 43 (GTX 1650) and Jetson Orin Nano (JetPack 6.x).
Validate Claims
here is a link to the demo:
https://github.com/vimalk78/dictate/?tab=readme-ov-file#demo
Specific Task(s)
Install dictate, start the daemon with dictate --serve &, then launch Claude with EDITOR=dictate-editor claude. Give Claude any task, then when you want to edit or expand on it, press Ctrl+G to open the voice editor. Press F5 to start recording, speak your prompt (e.g. "refactor this function to use async await"), press F6 to stop and transcribe, review the text, then :wq to send it to Claude.
Specific Prompt(s)
None
Additional Comments
No response
Recommendation Checklist
- I have checked that this resource hasn't already been submitted
- It has been over one week since the first public commit to the repo I am recommending
- All provided links are working and publicly accessible
- I do NOT have any other open issues in this repository
- I am primarily composed of human-y stuff and not electrical circuits