# SleekVoice privacy

> How SleekVoice handles voice on your Mac, and what this website records when analytics is configured.

Canonical: https://sleekvoice.app/privacy
Updated: November 20, 2026

SleekVoice is offline, local voice typing for macOS. The project states that Whisper runs on your Mac, including when you are offline, and that your voice stays on the machine. Transcription does not depend on sending your voice to a SleekVoice server. There is no SleekVoice account to sign into on this website, and the app does not publish a cloud history.

That is a product architecture claim, not a promise that the text you dictate stays private after it is typed. The transcript is inserted into whichever app is focused. If that app is email, Slack, a browser, or a document that syncs, the words leave your Mac because you put them there, the same way typed text would. Local speech-to-text does not encrypt the destination app.

## What the Mac app is documented to do

- Process speech with Whisper locally.
- Keep working when the Mac is offline.
- Type into the focused app after you hold the push-to-talk shortcut.
- Keep a per-app mode and a personal dictionary on the machine.

The application repository has not published a separate privacy policy, a data-retention schedule, or a list of on-disk files. This page will not invent those. If you need a control the app does not show, email [hello@sleekvoice.app](mailto:hello@sleekvoice.app).

macOS will ask before an app can use the microphone. Apple documents that switch at [Privacy & Security → Microphone](https://support.apple.com/guide/mac-help/control-access-to-the-microphone-on-mac-mchla1b1e1fe/mac). An app that types into other apps is commonly asked for Accessibility access, which Apple documents separately. Grant only what you intend to grant, and turn it off later in System Settings if you stop using SleekVoice.

## What this website does

This marketing site is static pages hosted on Vercel. It does not take dictation. It does not upload your voice.

If the environment variable `NEXT_PUBLIC_GA_MEASUREMENT_ID` is set, the site loads Google Analytics 4 from Google’s script. If that variable is empty, the analytics component renders nothing and no Google tag is added. The measurement ID is configured in Vercel. It is not written into the repository.

The site also includes Vercel Web Analytics and Vercel Speed Insights. Those record page views and performance measurements as described in Vercel’s own documentation: [Web Analytics](https://vercel.com/docs/analytics) and [Speed Insights](https://vercel.com/docs/speed-insights). They are part of the deployed site, not an optional package you toggle in the browser.

Pages you request are ordinary web requests. A hosting provider sees IP addresses and request logs the way any host does. This site does not run a comment system, a newsletter form, or a login.

## What this page does not claim

SleekVoice has not commissioned a security audit that this page can cite. It does not claim end-to-end encryption, a SOC report, or a HIPAA offering. Those phrases show up in other dictation products’ marketing. They are not SleekVoice claims.

Wispr Flow, by contrast, says on its [data-controls page](https://wisprflow.ai/data-controls) that transcription always occurs on the cloud, and it describes separate switches for model training and dictation cloud storage. That is Wispr’s description of Wispr. The [comparison](/compare/wispr-flow) quotes it and links it.

Questions about this page: [hello@sleekvoice.app](mailto:hello@sleekvoice.app).
