Whisper Dictate - Offline voice dictation that runs on your own GPU
by•
Push-to-talk voice dictation for Windows and macOS that runs entirely on your machine. Hold a hotkey, speak, release — auto-edited text lands at your cursor in any app: fillers removed, spoken punctuation applied, sentences capitalized. Say “scratch that” to rewrite mid-sentence, or use voice commands. English and German, powered by NVIDIA Nemotron models on your own GPU. No cloud, no audio leaving your machine. Source available; 14-day free trial, then €40/year.
Replies
Hi — Richard here, maker of Whisper Dictate. I built it because I wanted dictation that never leaves my machine: hold a hotkey, speak, release, and the text is pasted at your cursor, with audio and models staying entirely on-device (NVIDIA Nemotron streaming models run on your own GPU, English and German, with faster-whisper profiles as an option). It also does Wispr Flow-style auto-edits fully offline — filler words removed, "comma" and "new line" become real formatting, "scratch that" drops what you just said. v1.0.4 shipped today with a round of push-to-talk and model fixes for Windows and macOS. The app is source-available (public repo) and there's a free community edition with a smaller model set if you'd rather poke around first. Happy to answer anything about the on-device models, the hotkey plumbing, or German ASR quirks.