Doclin vs Napkin AI

Hold Right Command, speak naturally, and release to insert text into the field you were already using — editors, browsers, terminals and chat boxes alike.

The standard path uses Apple's on-device speech recognition and Kokoro/macOS voices, so your recordings and transcripts are not uploaded. Audio in local mode is never written to disk.

Hear progress from Codex (desktop and CLI) and Claude Code read aloud, so you can stay aware of a running task without watching the terminal.

OpenAI transcription, transcript cleanup, notification summaries and higher-quality notification voices are individually switchable and use your own API key.