Extensions connect another program to utt. They can send audio for transcription, read live or finished transcripts, change text before utt pastes it, and add settings or status to the utt window. Every extension needs your approval before utt shares anything with it. The developer guide explains how to build one.
What an extension can do
Choose only the features your extension needs.
Transcribe recordingsDrop an audio file into a local folder. utt writes the transcript beside it.
Receive finished transcriptsGet the original words and final text for each finished transcript.
Read words liveReceive an English preview while someone speaks. The final transcript arrives separately and can differ.
Change text before pastingLet your program edit or block a transcript before utt pastes it.
Add settings and statusPut your program's settings, actions, and live status in utt.
Show a message on screenShow a short message across every screen, just below the Dock. Any mouse or keyboard input dismisses it.
Offer text to copyShow text snippets with a Copy button on your extension's page. They stay available when your program is stopped.
Show the dictation yourselfStart a dictation that your program shows. utt hides its own panels, pastes nothing, and sends you the level, the live words, and the transcript.
TerminalAvailable
Dictate into OpenCode's prompt without leaving the terminal.
Press ctrl+s and speak. A level meter takes over OpenCode's footer, your words appear in the prompt as you talk, and utt replaces them with the finished transcript when you stop.
Choose a session on your phone, read its latest reply, and hold to talk. utt transcribes the recording on your Mac, then Deckhand sends the text to that session.
Give your coding assistant a transcript of any video on your Mac.
Ask your coding assistant to transcribe a video. Reel extracts the audio, splits long recordings at pauses, and sends each part to utt. Transcription stays on your Mac and uses your chosen model and text settings.
Deckhand VR will put your Claude Code sessions into panels around you. Choose a session and speak to it as you would in the iPhone app. Development has not started.