Privacy Policy

Speakeasy · last updated 13 September 2026

Short version: your talks live on your phone. The app keeps a recording of each talk on the device so you can hear it back, and nothing about you is uploaded unless you switch on one of two optional features, both described below, both off until you turn them on. Signing in is optional, and what the server holds if you do is one name and one date per talk.

Your voice

Speakeasy uses the microphone only while a drill is running. Your speech is turned into text by your own device, using Apple's on-device speech recognition. The app requires on-device transcription and will refuse to run rather than send your audio anywhere.

The app keeps a recording of each talk on your phone, so you can hear it back on the results screen and from History. Recordings are kept for your last sixty talks, are excluded from your device's backups, and are deleted with the talk they belong to. Audio is never uploaded by any feature of the app.

What is stored on your device

Your talks, meaning the two words, the transcript, the timed sentences it was made of, the pace, the pauses and the date, are saved in a single file on your own device. They are not uploaded and they are not shared. The file is included in your device's own iCloud or computer backup if you have that switched on, the same as any other app's data; the recordings are not.

Deleting the app deletes all of it. Settings also has a "Delete all history" button that removes every talk and every recording without uninstalling, and you can delete a single talk, with its recording, by long-pressing it in History.

Two small files the app reads on every launch

On every launch the app reads two small files published alongside it: a settings file that can switch a broken feature off from our end, and a content file that carries the word packs. The settings file is read whether or not any optional feature is on, because a switch that only reached the people who opted in could not take a broken build out of service for anybody else. They are the same files for everybody, they say nothing about you, and reading them sends nothing about you. They are fetched from GitHub Pages, which sees the request the way any web server sees a page load.

Reminders

If you turn on the daily reminder, it is scheduled by your own device. No push token is registered, nothing is sent to a server, and turning it off in Settings removes every pending reminder.

The home screen widget

The widget reads today's two words, which are computed on the device from the date, and your streak count, which is stored in your device's keychain so that the widget can see it. Neither leaves the phone.

Signing in, if you choose to

Signing in with Apple is optional and the app works fully without it. If you do, a server of ours holds the name you pick and the date of each drill you finish, and that is the whole record: no talk, no transcript, no recording, no result. Delete account in Settings removes that record and tells Apple to drop the sign-in too, so it disappears from your own Settings as well.

The connection check, if you switch it on

Speakeasy can check whether you actually linked the two words you were given, rather than just saying both. That is a judgment about meaning, so it cannot be worked out on your phone by counting, and it is one of the two features that send any part of a talk anywhere.

It is off until you turn it on, in Settings, behind a screen that says all of this before you agree. With it on:

• the text of your talk, and the two words, are sent to Google's Gemini API, through a Speakeasy server that holds the key so the key never ships inside the app. It is sent once while you are still speaking, so the answer is ready the moment you stop, and again at the end only if what you said moved on a long way
• Google's own terms and privacy policy govern what they do with that text
• your audio is never sent
• your name, your streak, your history and every other talk are never sent
• turning the switch off stops it immediately, and the rest of the results screen is unaffected

Your material, if you use it

With Speakeasy Pro you can give the app a document you have to speak about, a deck's notes, an essay, an interview brief, either as a file or as pasted text, and it writes a practice pack from it: word pairs in the document's own terms, the questions a listener would ask, positions that can be argued, and ideas to explain.

• the text of the document goes to Google's Gemini API once, through the same Speakeasy server, when you press Write my practice, and at no other time. A screen says this before the first one
• the app keeps the pack it wrote and not the document; the server stores neither
• Google's own terms and privacy policy govern what they do with that text
• Remove it, on the same screen, deletes the pack from your phone

No tracking

No analytics, no advertising identifiers, no third-party tracking SDKs, no crash reporting services, no cookies. Nothing about how you use the app is collected, profiled or transmitted, and nothing is sold or shared with anyone for advertising or any other purpose. Apart from the two files above and Apple's speech model, which the phone downloads once over an unmetered connection so that transcription can run on your own device, the app makes no network request unless you sign in or switch on one of the two features described above.

Withdrawing permission

You can revoke the app's access to the microphone or to speech recognition at any time in iOS Settings, under Privacy & Security, or by finding Speakeasy in the list of apps. The app will tell you it cannot run and will not attempt to work around it. Talks and recordings already on the phone are unaffected until you delete them.

How long things are kept

Recordings: on your device for your last sixty talks, then the oldest is removed. Transcripts and statistics: on your device until you delete them. A practice pack written from your material: on your device until you replace or remove it. The sign-in record, if you signed in: on our server until you delete your account. There is no expiry beyond these and no separate archive.

Children

Speakeasy is not directed at children under 13 and does not knowingly collect personal information from them. Signing in is optional, and the only thing collected from anyone who does is a chosen name and dates.

Purchases

Speakeasy Pro is an auto-renewing subscription handled entirely by Apple. Payment details are never seen by the app or by us; Apple tells the app only whether a subscription is active. Manage or cancel it in your Apple Account settings.

Changes

If this policy changes, the updated version will be posted at this address with a new date at the top.

Contact

Questions about this policy: help@vsvsllc.com