DictDrop is a focused tool — no subscription upsells, no feature-gated tiers. You buy it once and get everything.
DictDrop uses Groq's fastest speech recognition model — purpose-built for real-time inference, which is why the latency is dramatically lower than cloud APIs built on shared infrastructure. Most dictations complete in 200–350ms from when you stop speaking to when text appears.
Accuracy on technical vocabulary — variable names, framework names, domain jargon — is significantly better than Windows Speech Recognition or browser dictation.
Most "voice-to-text" tools copy the transcript to clipboard and send Ctrl+V. This overwrites whatever you had copied, creates a visible paste flash, and fails in rich text editors that intercept paste events.
DictDrop delivers text directly to wherever your cursor is — at the OS level, not the clipboard level. The receiving app cannot tell the difference from physical keyboard input.
DictDrop has no backend server. When you dictate, audio goes: your mic → Groq API (authenticated with your personal key) → transcript returned → injected at cursor. We are not in that chain at any step.
Your Groq account's free tier covers hundreds of hours of dictation per month at zero marginal cost. If you go heavy, Groq's pay-as-you-go is fractions of a cent per minute.
No DictDrop servers in this chain
V1.2 introduced a lightweight floating status dot — a small circle that shows recording state without creating a taskbar entry or stealing focus from your work. It runs silently in the background at all times.
Green = actively recording. White = idle. The dot appears at a fixed corner position you can configure in settings.json. At rest, it consumes zero CPU.
All features
One purchase. All of this. Today.
Groq's fastest inference — faster than any cloud GPU-based transcription service.
Bring your own Groq key. We are not in the audio pipeline. Zero data retention on our end.
Text lands at your cursor — no clipboard overwrite, no paste, no flicker. Indistinguishable from keyboard input.
Browser, Electron, native desktop, legacy apps — any window that accepts keyboard input.
Default Ctrl+Shift. Change it to any modifier combo in settings.json.
Lightweight floating dot. No taskbar entry. Runs silently — zero CPU at rest.
Fully self-contained — bundled runtime, zero dependencies to manage, no PATH changes needed.
All future version updates are included. No upgrade pricing, no annual renewal.
Uses your system default audio device. Works with headsets, USB mics, built-in mic — anything Windows recognises.
Roadmap
These are actively in design. Sign up at the bottom to get notified when they ship.
Status dot follows active window across monitors automatically.
Adjustable detection threshold — fewer false triggers in noisy environments.
Local speech recognition — transcription without any internet connection.
One email per release — no newsletters, no promotions.