Mevro Captions for iPhone and Android
Privacy Policy
Effective: 12 November 2026 (previous version: 9 November 2026)
For the iPhone and Android app Mevro: Auto Captions for Video. This is separate from the mevro.app web service, which has its own Privacy Policy.
This policy explains what personal data the Mevro apps for iPhone and Android ("the App", "we", "us") collect, how we handle it, what rights you have under the General Data Protection Regulation (Regulation (EU) 2016/679, "GDPR") and equivalent laws, and how to contact us about any of it.
The App is operated from Poland by the operator of mevro.app, who is the data controller. Contact: support@mevro.app.
This policy covers the Mevro iOS app (App Store) and the Mevro Android app (Google Play). Where the two differ, the text says "on iPhone" or "on Android". The Mevro web app at mevro.app has its own Privacy Policy.
The short version
- No account. You never sign up or sign in. We don't ask for your name, email or date of birth.
- Your videos stay on your phone. Videos, captions, imported music and logos are processed on your device and are never uploaded.
- Audio is transcribed by ElevenLabs. To write your captions, the App sends only the sound of your video (not the picture) to our speech-to-text provider, ElevenLabs, which returns the words and their timings. Right after that, the App asks ElevenLabs to delete the transcript from our account. Automatic captions therefore need an internet connection; the App has no on-device speech-to-text.
- We collect a small amount of usage and crash data (through Google Firebase) to fix bugs, understand which features are used, and test different free-tier offers.
- Apple or Google handles payments. We never see your card, Apple ID or Google account details.
- We measure our own App Store ads. On iPhone the App asks Apple whether your install followed one of our ads. The answer names one of our campaigns. It does not name you or your phone.
- We do not sell your data, show ads, or track you across other companies' apps and websites.
1. What stays on your device
The following is created, stored and processed only on your phone, inside the App's private storage. It is never sent to us or to any third party by the App (the one exception — the audio track sent for transcription — is described in section 2):
- Videos you import or record. The picture never leaves your phone.
- Captions and your edits to them. The transcript ElevenLabs returns is stored only on your phone.
- Edits and AI tools. Cutting silences and filler words, auto zoom, auto emojis, keyword highlights, hook titles and audio clean-up all run on your device. Auto zoom finds faces in the frame on the device — with Apple's Vision framework on iPhone and Google ML Kit face detection on Android; no image or face data is stored or sent anywhere. Where your phone supports it, keyword, emoji and hook-title suggestions use an on-device language model: Apple's Foundation Models on iPhone, Gemini Nano (through Android's AICore) on Android. Caption translation runs on the device with Apple's Translation framework on iPhone and Google ML Kit Translation on Android.
- Music you import from your files and logos you add to your brand kit.
- Your caption styles, brand kit and settings.
- Exported videos and
.srtfiles, until you save or share them.
App files are excluded from backups: from iCloud backup on iPhone; on Android the App turns off Google backup and device-to-device transfer. Deleting a project deletes its files; deleting the App deletes everything above. To apply the free-tier export limit, the App stores the dates of your recent free exports in secure storage on your device — the iOS Keychain on iPhone, which iOS may keep after the App is deleted, and storage encrypted with the Android Keystore on Android, which is removed when you uninstall the App. These dates contain no personal content.
When you export or share. Saving writes the video into your photo library (Photos on iPhone, your gallery on Android). Sharing to TikTok, Instagram, YouTube or any other app only happens when you tap a share button; from that point the receiving app's own terms and privacy policy apply.
2. What we collect and why
Transcription (ElevenLabs)
When you caption a video, the App extracts the video's audio track (compressed, speech quality) and uploads it directly to ElevenLabs (ElevenLabs Inc. and its affiliates, elevenlabs.io), which transcribes it with its Scribe speech-to-text model and returns the text with word timings and the detected language. Only the audio is sent — never the video picture, file name, project name or any account or contact details. The audio contains whatever is said in your video, so it can include voices and personal information about you or other people.
Before each upload the App asks our server (a Google Cloud Function in the EU, region europe-west1) for a one-time upload permission. That request carries a random installation identifier created by the App, the length of the audio, and a Firebase App Check token that proves the request comes from a genuine copy of the App (issued using Apple App Attest on iPhone and Google Play Integrity on Android — see "Device integrity checks" below). We use these only to prevent abuse and keep a daily limit per installation: our server stores a hashed form of the identifier with a daily count of transcriptions and audio minutes, which is deleted automatically after 7 days. The audio itself never passes through our server.
Right after the transcript comes back, the App asks our server to delete it at ElevenLabs. That request carries the same random installation identifier, a Firebase App Check token and ElevenLabs' reference number for the transcript (an opaque ID, not its text). Our server forwards the deletion to ElevenLabs. So that this can't be misused, an installation can only delete as many transcripts as it received upload permissions for: our server keeps that count on the installation's daily usage record and stores a hashed form of the reference number (both deleted automatically after 7 days), and records only a shortened hash of the reference number in its logs. If the deletion can't be sent — for example because you went offline — the App keeps only the reference number on your phone and retries the next times you open it, for up to five failed attempts and at most 7 days.
Purpose: turning speech into captions. Legal basis: performance of contract — it is the service you ask for (art. 6(1)(b) GDPR) — and, for the abuse limits, legitimate interests in protecting the service (art. 6(1)(f)). ElevenLabs processes the audio on our behalf as a data processor under its data-processing terms.
What ElevenLabs keeps. Your audio is sent to ElevenLabs only to create the transcript. By default ElevenLabs stores the transcript of each request in our ElevenLabs account; right after the transcript comes back, the App asks ElevenLabs to delete it, and once deleted it can no longer be retrieved from our account. ElevenLabs offers no separate way for us to delete the uploaded audio, and does not state whether deleting the transcript also removes the stored audio, so ElevenLabs may keep the audio — and copies of the audio or transcript in its backups — for a limited time under its own retention policy, to provide, secure and troubleshoot its service. Our plan does not include ElevenLabs' "zero retention" mode (it is offered to enterprise customers only). We have opted out of ElevenLabs using our data to train its models.
Automatic captions are only available this way: the App has no on-device speech-to-text. If you don't want a video's audio to leave your phone, don't use automatic captions for it. When your phone is offline, or the transcription service can't be reached, the App shows an error and sends nothing; you can try again later.
Usage analytics (Google Firebase Analytics)
Which screens you open and which features you use (for example: a video was imported, captions were generated, a style was applied, an export finished or failed, the paywall was shown or a trial started, or the star rating you gave in the feedback survey), plus coarse details such as the length bucket of a video, the detected spoken language, which free-tier offer you were assigned and whether you have Pro. Firebase also records a random app-instance identifier, your device model, operating-system version, App version, language/region settings and an approximate location derived from your IP address (country/region level). The App does not collect the advertising ID (IDFA on iPhone, Android advertising ID on Android).
We never log transcript text, video or file names, project names, or the content of your videos. Used to understand how the App is used and to improve it. Legal basis: legitimate interests — understanding and improving the App (art. 6(1)(f) GDPR). You can switch it off at any time in Settings → Share usage data.
Ad measurement (Apple Ads on iPhone, store links on Android)
We advertise the App on the App Store with Apple Ads. To learn which of our ads work, the App on iPhone asks Apple, once after installation, whether the install followed one of our ads. Apple answers with the numbers of our own campaign, ad group, keyword and ad, the country or region of the store, and whether it was a first download or a re-download. The answer contains no identifier of you or of your device. The App does not use the advertising identifier (IDFA) and does not ask for tracking permission.
That answer is stored with your usage analytics in Google Firebase, so that we can see, for example, how many people who came from an ad finish a video. Firebase Analytics and RevenueCat each ask Apple the same question for their own reports: Firebase to show installs per ad keyword, RevenueCat to count trials and purchases per campaign. On Android, Google Play tells Firebase which store link led to the install when that link carries a campaign name of ours.
So that subscription events (a trial started, a renewal, a cancellation) appear in the same analytics, RevenueCat receives the random Firebase app-instance identifier described above, on iPhone and on Android, and reports those events to Google Analytics under it.
Legal basis: legitimate interests — measuring our own advertising (art. 6(1)(f) GDPR). None of the above happens while Settings → Share usage data is off.
Feedback survey and app ratings
After your first export the App asks once "How was your first video?". If you answer, the star rating (1–5) and whether you typed a comment are sent as a usage-analytics event (above, same switch); declining is recorded the same way. The comment itself is not sent by the App: if you type one, the App opens your own email app with a message to support@mevro.app containing the comment, the App version and your platform (iPhone or Android), and you decide whether to send it. If you do, we use your email and its contents only to answer and to improve the App (legitimate interests, art. 6(1)(f) GDPR) and delete it when it is no longer needed for that.
From your second export the App may also ask iOS or Android to show the standard "rate this app" dialog. Apple or Google decides whether it appears; a rating or review you leave there is handled by Apple or Google under their own policies, and we only see what is published on the store. Your answer to our survey has no effect on whether this dialog is requested.
Crash reports (Google Firebase Crashlytics)
When the App crashes or hits a serious error: the stack trace, device model, operating-system version, App version, free memory/disk state at the time, and a random installation identifier. Each report also carries a short log of the App screens and steps that led up to it, your plan (Free or Pro) and free-tier offer, and the settings of your last export (format, quality, length range) — never your videos, captions, transcripts, file or project names. Used to find and fix bugs. Legal basis: legitimate interests — keeping the App reliable. Crash reports follow the same Share usage data switch.
Offer testing (Google Firebase Remote Config and A/B Testing)
The App fetches settings from Firebase Remote Config, using a random installation identifier. We use this to test different free-tier offers (for example, how many watermarked exports are free) and to measure which works best, combined with the usage analytics above. Legal basis: legitimate interests — improving the App and its pricing.
Purchases (Apple, Google Play and RevenueCat)
Subscriptions are sold and charged by Apple through the App Store on iPhone and by Google through Google Play on Android. We never receive your card number, billing address, Apple ID or Google account details. We use RevenueCat to check whether you have an active subscription: RevenueCat receives the store's transaction details for your purchase (product, price, currency, dates, trial and renewal status, and the store's purchase token or receipt), an anonymous identifier it generates for your installation, and basic device information (such as operating-system version, country/storefront and IP address). Used to unlock Pro and to restore purchases. Legal basis: performance of contract (art. 6(1)(b) GDPR) and compliance with tax and accounting obligations (art. 6(1)(c) GDPR).
Device integrity checks (Apple App Attest, Google Play Integrity)
To protect the transcription service from abuse, Firebase App Check asks the platform to confirm that the request comes from a genuine, unmodified copy of the App on a real device: Apple App Attest on iPhone, the Google Play Integrity API on Android. Apple or Google evaluates signals about the App and the device and returns a verdict; we receive only a short-lived token, not those signals. Legal basis: legitimate interests — protecting the service (art. 6(1)(f) GDPR).
Apple and Google on-device services
- Apple (iPhone). Apple's Translation framework may download language files from Apple when you first translate into a new language (iOS asks you first). This is handled by iOS under Apple's Privacy Policy; we receive nothing from it.
- Google ML Kit (Android). Face detection (auto zoom) and caption translation use Google's ML Kit libraries, which run on your phone. When you first translate into a new language, the App asks you before ML Kit downloads that language's model (about 30 MB) from Google. ML Kit may also contact Google to send API usage and performance diagnostics (for example, which ML Kit feature ran, how long it took, the device model and Android version) and to receive model updates. No images, faces, audio, captions or translated text are sent. Google handles this under the Google Privacy Policy and the ML Kit terms.
- Gemini Nano (Android). On phones that support it, suggestions use Gemini Nano through Android's AICore system service; the text is processed on the device and isn't sent to Google or to us. AICore and Google Play services download and update the model under Google's policies.
3. Permissions the App asks for
On iPhone:
| Permission | Why | If you say no |
|---|---|---|
| Photos (read) | To show your videos so you can pick one to caption, and to pick a logo for your brand kit. You can allow only selected items. | You can still import videos from Files. |
| Photos (add) | To save your captioned videos to your photo library. | You can still share the exported file with “More”. |
| Camera | To record a new video to caption. | Use Photos or Files instead. |
| Microphone | To record sound with that video so it can be captioned. | Use Photos or Files instead. |
You can change these at any time in iOS Settings > Mevro.
On Android:
| Permission | Why | If you say no |
|---|---|---|
| Notifications (Android 13 and newer) | To show the progress of a caption or export job that keeps running while you use other apps, and to tell you when it’s done. | Jobs still run; you just don’t see the notification. |
| Storage (Android 9 and older only) | To save your captioned videos to your gallery. Newer Android versions need no permission for this. | You can still share the exported file. |
On Android the App never asks to read your photos, videos or files: you pick a video or logo with Android's system photo picker or file picker, and the App only receives the item you chose. Recording a video opens your phone's camera app, so the App itself does not ask for camera or microphone access. You can change permissions at any time in Android Settings > Apps > Mevro.
On both platforms the App does not use your precise location, your contacts or the advertising identifier, and does not track you.
4. How long we keep it
- On-device data: until you delete the project or the App (see section 1).
- Audio sent for transcription: we keep none. The App asks ElevenLabs to delete each transcript from our account right after it arrives (and retries on later launches if that fails). ElevenLabs may keep the audio, and backup copies, for a limited time under its own retention policy (see section 2).
- Transcription limit counters (hashed installation id, daily counts of transcriptions and of deletions they allow, hashed transcript reference numbers): deleted automatically after 7 days.
- Pending transcript deletions: a failed deletion's ElevenLabs reference number stays on your phone until the deletion succeeds, has failed five times, or 7 days have passed.
- Firebase Analytics: event-level data is kept for 14 months, then deleted; aggregated reports may be kept longer. Our own copy of the event data (Google BigQuery, in the EU) is deleted after 14 months too.
- Firebase Crashlytics: crash reports are kept for up to 90 days.
- RevenueCat: purchase records are kept for as long as needed to provide your subscription and restore it, and for at least 5 years where Polish tax law (Ordynacja podatkowa) requires records of sales. Apple and Google keep their own records under their policies.
5. Who we share with (sub-processors)
Each holds only the data needed for its function:
- ElevenLabs — United States and other regions — transcribes the audio track of the videos you caption (speech-to-text), and deletes each transcript from our account when the App asks.
- Google (Firebase Analytics, Crashlytics, Remote Config, A/B Testing, App Check, Cloud Functions, Cloud Firestore, BigQuery; on Android also Play Integrity and ML Kit) — United States, the EU and other regions — usage analytics, crash reports, offer testing, device integrity checks, the transcription upload permission, transcript deletion requests and daily limits, and ML Kit's diagnostics and model downloads.
- RevenueCat — United States — subscription status, receipt validation, and counting purchases per ad campaign.
- Apple (Apple Ads) — tells the App on iPhone whether an install followed one of our ads (Apple is an independent controller for its advertising service, under its own privacy policy).
- Apple (iPhone) and Google Play (Android) — payments, subscriptions, refunds and app distribution (each is an independent controller for this, under its own privacy policy).
We do not sell your personal data. We do not share it with advertisers. We only disclose data to authorities when legally compelled by a valid, jurisdictionally proper request.
6. International transfers
ElevenLabs, Google and RevenueCat may process data outside the European Economic Area, primarily in the United States. Transfers are covered by Standard Contractual Clauses (2021/914/EU) and, for US recipients certified under it, by the EU–US Data Privacy Framework.
7. Your rights under GDPR
You can exercise the following rights at any time, free of charge: access, rectification, erasure, restriction, portability, objection to processing based on legitimate interests (analytics, crash reports, offer testing), and withdrawing consent where we rely on it. You can also lodge a complaint with a supervisory authority — in Poland, Urząd Ochrony Danych Osobowych (uodo.gov.pl).
Because the App has no accounts, the analytics, crash and transcription-limit data we hold are tied only to random identifiers, not to your name or email, so we usually cannot find "your" data from an email alone. The simplest way to stop all collection and start fresh is to delete the App: this deletes the random identifiers on your device. You can also turn off usage analytics at any time in the App under Settings → Share usage data; once it is off, no further analytics events are sent. Crash reports follow the same switch. To exercise any other right, email support@mevro.app and we will help — for purchase data, including the date of purchase lets us find it. We respond within one month.
8. Tracking and cookies
The App does not track you across apps or websites owned by other companies, does not use the advertising identifier, and shows no ads. We measure our own App Store ads as described in section 2; that uses Apple's own record of its ads, which identifies neither you nor your device. The App does not use cookies. Firebase and RevenueCat store random identifiers on your device so that their services work, as described in section 2.
9. Children
The App is not intended for children under 13 (or under 16 in countries that set a higher age of digital consent under GDPR art. 8). We do not knowingly collect personal data from children. If you believe a child has used the App and we hold data about them, email us and we will delete what we can identify.
10. Security
Your videos never leave your device through the App, so they are protected by your phone's own security (screen lock and device encryption). The audio sent for transcription goes straight to ElevenLabs over an encrypted connection (HTTPS) using a single-use permission that expires after 15 minutes; our ElevenLabs key never ships in the App. Data sent to Firebase, RevenueCat and Google also travels over encrypted connections (HTTPS).
11. Your content and other people's rights
You are responsible for having the right to use the videos, music, logos and other material you bring into the App, and for having the consent of people who appear in your videos where the law requires it. See the Terms of Use.
12. Changes
We may update this policy as the App evolves and will note the new effective date above. Material changes — those that expand what we collect or who we share with — will be announced in the App or in the App Store / Google Play release notes before they take effect.
13. Contact
Questions or data requests: support@mevro.app.
Copyright or abuse notices: abuse@mevro.app.