Dettivo Linux — AI skill for Claude Code
Private, offline speech-to-text for Linux: push-to-talk dictation into any app, meeting transcription with speaker diarization, and searchable history.
How to install Dettivo Linux
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open gmickel/dettivo-linux and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Dettivo Linux does
Private, offline speech-to-text for Linux: push-to-talk dictation into any app, meeting transcription with speaker diarization, and searchable history. Whisper, Parakeet and a local LLM on your GPU (Vulkan) or CPU. Native Omarchy and Hyprland integration, Wayland, MCP for AI agents.
Alternatives in AI
- Azure OpenAI .net — GPT-4, embeddings, DALL-E, and Whisper client 1.8k ★
- AI Avatar System — 🎭 AI Avatar / digital human platform — upload a photo, clone a voice, talk to any face in real time with lip 422 ★
- Yt Transcriber — CLI app- Give it a YouTube URL and you get a transcription with possible speaker identification and optional s 337 ★
README
Dettivo for Linux
Hold `F9`, talk, let go, and your words land in the window you were typing in. Dettivo is a speech workstation for the Linux desktop. It handles dictation, meeting transcription with speaker names, and a searchable history of everything you said, and it runs every model on your own machine. Audio and text leave the machine only when you point Dettivo at an endpoint yourself.
It is built for Omarchy and Hyprland first. It lives in the Omarchy bar, follows the active theme on every surface, and takes the `F9` chords Omarchy already binds for dictation. It also runs on Sway, Niri and any desktop whose portal offers global shortcuts.

What you get
- Dictation into any app. Hold to talk or toggle a longer take. Text goes in through the Wayland virtual keyboard, or a paste where the window refuses it, and always to the window that had focus when you pressed the key.
- Three modes. Raw keeps the engine's words, with file names, paths and URLs kept byte for byte. Polish fixes grammar, fillers and punctuation with a preset per app. Enhanced adds a rewrite by a local language model and falls back to the polished text if the model does not answer in time.
- Meetings without a bot. Your microphone and the call's audio are recorded as two tracks on one clock and transcribed live. At stop Dettivo builds the full transcript, names the speakers, and writes a summary, decisions and action items beside your notes. Exports go to md, txt, srt, vtt or json. A headset that drops, a daemon crash or a reboot leaves a meeting you can recover.
- History you can search. Every dictation and meeting is kept with its audio, full-text searchable, re-runnable with another model and exportable.
- Agents can use it. One command registers Dettivo as an MCP server in Claude Code, Codex, Cursor or Claude Desktop, with nineteen tools to search transcrip
Related Skills
Shiyin
macOS meeting notes: live transcript, local transcription (Breeze ASR / whisper.cpp), speaker diarization, AI
Talkshot
Push-to-talk visual notes for macOS - screenshot + voice note capture with on-device transcription, built for
Whisperx Transcribe
Claude Code skill/plugin: transcribe audio & video into clean, speaker-labeled Markdown via WhisperX, so an LL
Whisperflow Local
A 100% local WhisperFlow clone for Windows — Ctrl+Win to dictate, faster-whisper (CUDA) + a local Ollama clean
Localvoxtral
Talk to your coding agents by voice. Realtime, fully local macOS dictation that streams words as you speak and
Video To Text
Turns YouTube and Twitter/X videos into readable articles (PT-BR + English) — local transcription via Whisper,
Related Agents
Voice Engineer
GAIA voice interaction specialist. Use PROACTIVELY for Whisper ASR, Kokoro TTS, the Talk SDK, speech-to-speech
Talk Builder
Use to turn an accepted paper into a conference talk: a deck on the lab gabarit with timed speaker notes, in P
Personal Ingest
Ingests personal/private material — meeting notes (Granola, pasted, hand-summarized), voice memos (.opus / .m4