AI DICTATION
Speak into your microphone, capture your computer's own audio, or transcribe a recording — and watch the words appear. With an on-device model, everything runs on your own machine, so your audio never has to leave it.
Get it on Microsoft Store See what it does ↓
Free to start • On Windows • The on-device engine works offline
Dictate, capture system audio, or transcribe a file — and get clean, editable text with timestamps and subtitles, all on your own device.
Dictate live from your microphone, transcribe your computer's own audio — a meeting or a video that's playing — or turn an existing recording into text. Switch sources in a click, and a live meter shows your input is being heard.
With the on-device Whisper engine, your audio is transcribed entirely on your device — turn the network off and it keeps working. Transcripts and recordings are encrypted at rest and never uploaded. Cloud transcription is opt-in and clearly marked.
Save your transcript as plain text, or as SubRip (.srt) and WebVTT (.vtt) subtitles to caption a video — turn on timestamps for accurate cue timing — or as JSON with every segment's timing for your own tools.
Every recording is saved, so you can play it back with a waveform scrubber and watch each word highlight as it plays. Re-run a recording through a more accurate model any time — the result is saved as a new entry, never overwriting the original.
Transcribe dozens of spoken languages — choose one or let it auto-detect. The interface itself is available in 16 languages and follows your operating system automatically, with a one-click switch in Settings.
Every transcript is saved locally and encrypted, ready to reopen, copy or continue. Search any transcript with find — which also filters your recent list to recordings that mention what you typed — and delete any entry whenever you like.
Speech models also mark non-speech sounds like [Music] or (laughs). Choose how they're handled — keep them, strip the markers but keep the words, or drop whole music and noise lines — with the raw model output one panel away.
Start free with a capable on-device Whisper model — it is selected and downloaded automatically the first time you record. Pro unlocks larger models, the OpenAI cloud engine, and a GPU-capable local engine for the toughest audio.
Stop finishes your last words instead of dropping them, copy or save in a click, and drive everything from remappable keyboard shortcuts. Keep models loaded so repeat transcriptions start instantly, with built-in GPU safety limits for long jobs.
Choose your microphone, your computer's audio, or an audio file. The free on-device engine downloads once on first use, then runs offline — nothing is sent anywhere.
Click Record and start speaking, or choose a file and press Transcribe. The text streams in segment by segment as you go.
Copy the transcript, save it as text or subtitles, replay the recording, or re-transcribe it with a better model. Every take is kept in your encrypted local history.
When the recording is sensitive or simply yours to keep, AI Dictation brings the AI to your audio instead of sending your audio to the cloud.
Interviews, meetings, medical notes and legal dictation stay on the device — a fit for healthcare, legal, finance, government and other settings where audio can't go to the cloud.
No per-minute cloud transcription charges. Transcribe as much as you like with the free on-device engine, and keep working offline.
Turn a meeting into minutes, a voice memo into text, or a video into subtitles — in minutes, without leaving your desktop.
Choose which engine runs, keep your audio on your own machine, and automate transcription from the command line or an MCP-connected agent — without a third party in the loop.
Get AI Dictation now from the Microsoft Store on Windows, with a headless CLI and MCP server for automation. Editions for macOS, Linux, browser and mobile are on the roadmap.
Free gives you the complete app — live dictation, system-audio and file transcription, subtitle export, playback, re-transcribe, encrypted history, find and the CLI — with a capable on-device Whisper model and no time limit or nag screens. Pro simply unlocks larger models, the OpenAI cloud engine and a GPU-capable local engine when you want extra accuracy; no feature is ever locked away.
Students, educators, researchers, non-profits and other qualified users can apply for a free 12-month Pro licence.
Free to start, on-device by design. Get AI Dictation on Windows from the Microsoft Store, or read the plans first.