Open-Less/openless

Hold a hotkey, speak, release — polished text lands wherever your cursor sits. Not a dictation tool, not a clipboard hack, but a system-wide voice input layer that rewrites your speech into clean prose before it hits the screen. The differentiator is the polish step: it is not transcription dumped raw into your app, it is transcription run through an LLM pass that strips the ums, fixes grammar, and matches register. Works in any app on macOS and Windows without plugins or browser extensions. Open-source, so you can wire it to whatever model backend fits your threat model — local or cloud. The obvious reservation: latency on that polish step depends entirely on what model you point it at, so a slow API key turns a slick UX into an annoying pause. Worth thirty minutes to benchmark against your usual writing apps. -> Best for: solo founder or indie hacker who writes a lot and is tired of context-switching to clean up voice notes