How can we help?

Getting started

1. Open Speak Easy Privately and complete the first-run checklist.
2. Allow microphone access so the app can hear you while recording.
3. In Lite, download the required Parakeet speech model. The full edition includes it.
4. Place the cursor in a text field and use your configured shortcut, or choose Start Dictation from the app menu.

macOS permissions

Microphone

The microphone stays off between dictations. It opens only after you explicitly start recording and closes when you stop; Speak Easy Privately does not capture microphone audio in the background.

Keyboard Shortcut / Input Monitoring

Lets the app observe only your configured global shortcut while another app is active.

Automatic Paste

Optional. It lets Speak Easy Privately paste the finished text into the active field. Without it, the result remains available for manual paste.

If macOS asks you to quit and reopen the app after granting a permission, follow that system prompt once so the new decision can take effect.

For a plain-language explanation of the system-enforced macOS App Sandbox and permission boundaries, read Security by design.

Local speech and writing models

Full includes Parakeet, SAGE, Gemma and Qwen Fast and works offline from first launch. Qwen Balanced and the larger Qwen model are optional downloads.

Lite is a smaller download. Both website editions offer optional model downloads in the Speech Models tab, with progress, stop and retry controls. Each verified model is marked Ready when installation finishes.

Parakeet TDT 0.6B v3 is the supported speech-recognition model. Optional SAGE and Qwen models can improve punctuation and capitalization without sending text to a server.

Gemma 4 E2B is the recommended writing model, included in Full and available to download in Lite. Qwen Fast is a smaller alternative. Existing available model choices are preserved. If a previously bundled model is missing after an update, Full selects its included Gemma and explains how to download the previous model again. Text refinement remains optional.

Using dictation

• In the website editions, hold your configured shortcut to speak and release it to stop. Caps Lock and double-Option shortcuts toggle recording instead. You can also use the on-screen Record and Stop buttons.
• Watch the floating waveform for recording and processing status.
• Choose languages in Settings, including Automatic for mixed-language speech.
• Keep Smart Local correction off for raw ASR output, or enable it for conservative local punctuation.
• Local text history is optional and can be disabled or automatically deleted.

From version 1.6.6, Process long dictations while recording is enabled unless you previously saved a different choice. It processes your audio locally during an active recording to reduce the work remaining after Stop. You can turn it off in Settings; wording and punctuation may differ from processing the whole recording at once.

Insert text while recording is a separate, optional setting and is off by default. In the website edition, it requires Automatic Paste permission and processing while recording. Keep the same app, field and cursor selected, and avoid editing until Stop. Phrases appear after suitable pauses and processing, not at a fixed interval.

Troubleshooting

The shortcut does not respond

Open Settings → Permissions and confirm Keyboard Shortcut permission. Try a standard shortcut such as ⌘⇧D to separate Fn-key hardware behavior from permission setup.

Text is not pasted

Confirm Automatic Paste permission, then quit and reopen if macOS has just changed the setting. You can always paste the result manually.

Lite says a model is not ready

Open Speech Models, keep the Mac awake until download and verification complete, then confirm Parakeet shows Ready. A model activates only after its integrity check succeeds.

Contact support

For information about data handling, read the Privacy Policy.