Whisperly

macOS 1.0.1 · Windows 0.6.1

Hold a key. Speak. Paste.

Push-to-talk dictation for Mac and Windows. Press a key, talk, release — the transcript appears at your cursor in about 200 ms. Nothing leaves your machine.

Download for Mac Get the Windows preview

Free forever. Pro is a one-time $29. No subscription, no account.

The HUD floats near your cursor — hold Right ⌘ (Mac) or Right Ctrl (Windows), speak, release.

Three steps. Two seconds.

One hotkey, one local server, one paste. The whole loop lives on your machine.

  1. 1

    You press the key.

    Pick any modifier — Right ⌘, Right ⌥, fn, or a custom hyper-key. Whisperly records while it’s down and stops the instant you release.

  2. 2

    Whisperly transcribes locally.

    Audio flows into a warm whisper.cpp server pinned to 127.0.0.1. Typical round-trip: ~200 ms.

  3. 3

    Text lands at your cursor.

    Smart formatting, custom dictionary, and voice commands run before the transcript is pasted — into whatever app you happen to be in.

Built for daily use.

Four ideas Whisperly is built around — the rest follows from these.

Local-first by design.

Audio, transcripts, and telemetry never leave your machine. Whisperly reaches the network only to fetch what you ask it to install — the speech models and the engine — and for an optional once-a-day check for updates.

Warm and fast.

A persistent whisper.cpp server keeps your model in memory — each dictation finishes in ~200 ms instead of the second-and- a-half a cold start would cost.

Native to the platform.

Liquid Glass on macOS 26 with a clean fallback to system materials on 14 and 15. WPF + .NET 9 on Windows. SF Symbols and SF Pro throughout.

Yours to shape.

Pick a model from tiny to large-v3-turbo. Add words to the dictionary. Bind voice commands. Choose what gets formatted and what stays raw.

“Nothing you say is ever sent anywhere. Whisperly downloads the model you choose, then never sends your audio or text off the machine — not once, not anonymised, not for diagnostics.

— from the source code, not the marketing

Two ways to use it.

The free tier is fully functional. Pro adds the bigger models and the power-user surface.

Free

$0forever

Everything you need to dictate at your cursor.

  • Tiny, base, small whisper models
  • Push-to-talk hotkey with HUD
  • Local-only transcription
  • Transcription history + menu access
  • whisperly CLI
Get Whisperly

Pro

$29one-time

Everything above, plus the higher-end models and the power-user surface.

  • Medium & large-v3-turbo models
  • Custom dictionary (auto-corrections)
  • Voice commands (“new line”, “period”, …)
  • Spotlight-style history picker
  • Priority email support
Get Pro

Pricing is enforced inside the app. Buying Pro is a one-time unlock — no recurring charge, no account.

Download Whisperly.

Stable on Mac. Preview on Windows. Pick the right one and you’re dictating in under a minute.

For Mac

Whisperly 1.0.1 · macOS 14+ · Apple Silicon

Download Whisperly.dmg
  1. Open the disk image and drag Whisperly into the Applications shortcut.
  2. Launch — onboarding walks through Microphone & Accessibility.
  3. Pick a model, choose your hotkey, start dictating.

Ad-hoc signed (no paid Developer ID yet). First launch will show a Gatekeeper warning — right-click → Open.

For Windows Preview

Whisperly 0.6.1 · Windows 10/11 · per-user install

Download WhisperlySetup.exe
  1. Run the installer — per-user, no UAC. Lands in %LOCALAPPDATA%\Programs\Whisperly.
  2. SmartScreen warns it’s unsigned — click More info → Run anyway.
  3. Walk through onboarding and pick a hotkey.

Confirmed working on real Windows hardware. GPU transcription, setup wizard, prompt mode, history and file transcription.