Rekord AI transcribes speech to text entirely on your device. No cloud. No accounts. No data leaving your machine. Press a key — speak — done.
How it works
No setup wizards. No accounts. No cloud pairing. Just install, press a key, and start dictating.
Press and hold Numpad 0 (or any key you choose). Rekord starts listening immediately — no delay, no wake word.
Talk at a normal pace. Rekord captures your voice and processes it entirely on your machine using the Whisper model.
Let go of the hotkey. Your speech is transcribed and injected at your cursor in under a second. Clean, accurate text.
Text lands exactly where your cursor was — browser, VS Code, Outlook, CRM, chat. No paste, no copy, nothing to do.
Features
The free tier is genuinely useful. Pro removes all limits and adds the settings that power users want.
Buy once. No subscription.
Privacy
No microphone stream is sent to a server. No transcript is stored online. The Whisper model runs on your CPU, processes audio in memory, and discards it. Nothing is logged. Nothing is transmitted.
Every word is transcribed on your machine. We don't have servers listening to you.
The free tier requires nothing. Pro requires a license key — no profile, no login.
Rekord doesn't track usage, collect crash reports, or phone home for any reason.
Use cases
Anywhere you type, Rekord works faster.
Dictate full email drafts hands-free while reviewing other work.
Log call notes directly into your CRM in real time, no typing.
Slam out Slack, Teams, or WhatsApp messages at speaking speed.
Capture decisions and action items mid-meeting without looking away.
Draft blogs, scripts, and social posts 3x faster than typing.
Dictate comments, docstrings, and commit messages straight into your editor.
Comparison
Cloud tools like Wispr Flow and Glaido charge monthly and process your audio on their servers. Rekord doesn't.
Pricing
Free is real. Pro is a single payment — no recurring charges, ever.
No credit card. No email. Just download.
One payment. Yours forever.
One-time purchase — no subscription
Secure checkout via Lemon Squeezy. License key emailed instantly.
FAQ
Yes. All audio processing happens locally using the open-source Whisper model via faster-whisper. You can inspect the source code on GitHub. No network requests are made during transcription — you can run Rekord with your firewall fully closed after the one-time model download.
Only once — the first time you run Rekord, it downloads the Whisper model (~75–500MB depending on quality). After that, it works completely offline.
You purchase once via Lemon Squeezy. You receive a license key by email. Open Rekord, click Upgrade, paste the key — done. It activates locally and never needs to phone home again.
Windows 10 and Windows 11 (64-bit). Mac and Linux support is on the roadmap.
Very good for clear speech in English. The base model (free tier) handles everyday dictation well. The small and medium models (Pro) are noticeably more accurate for technical language and accents. All models improve with clear audio and a decent microphone.
Your purchase receipt includes a Lemon Squeezy customer portal link where you can retrieve your key at any time. You don't need to contact us.
Download Rekord, run the installer, hold Numpad 0 — and your first transcription happens before you finish reading this sentence.
Windows 10/11 · No account required · 100% local