Stet

Stet provides a native macOS voice input method that converts spoken words into text directly at the cursor position in any application. It captures speech locally on the Mac, allowing users to dictate up to four times faster than typing while preserving their natural speaking style, including filler words and informal phrasing. The system maintains a personal dictionary owned by the user and offers optional AI‑driven refinement through a cloud service or a personal API key, with the core transcription remaining on‑device for privacy.

The tool is aimed at individuals who want a seamless, system‑wide dictation experience without sacrificing control over their language data. It supports a broad range of desktop apps—from email and messaging to code editors and design tools—so users can dictate in the environments they already use. Stet’s design emphasizes minimal editing after speech, keeping the output true to the speaker’s voice while optionally providing AI‑assisted rewriting for more formal contexts.

Distinctive aspects include on‑device speech processing, a portable personal dictionary, and an open‑source implementation that can be extended or self‑hosted. A paid cloud option supplies unlimited AI refinement with one‑click activation, while a free tier lets users bring their own API keys for unlimited usage under full data privacy.