Privacy Policy
Effective 2026-09-10. Last updated 2026-09-21.
Your recordings stay on your iPhone. Recording, cloning and voice conversion all run on-device, and there is no account to create. The only thing that ever leaves your device is a voice you deliberately choose to publish to the Community Library — and even then Mimic sends a mathematical voice print, never your audio. Publishing attaches a random creator code so blocking works. If you never publish, no voice and no creator code is ever sent. Separately, Mimic counts which features get used — anonymously, with no identifier of any kind.
What stays on your device
Unless you publish a voice (see below), everything Mimic captures or generates lives in the app's local storage on your iPhone. That includes:
- Voice recordings — both the source audio you record into Mimic and any converted output saved in the Recordings tab. Audio is never uploaded, under any circumstances.
- Voice avatars and cloned embeddings — when you clone a voice, Mimic extracts a small mathematical fingerprint (a 256-number speaker embedding) from your sample and stores it locally. The original recording stays local too; you can delete either at any time.
- Voice conversion — both live (Performer Mode) and offline conversion run on-device through Core ML on Apple's Neural Engine. No audio is uploaded for processing.
- Pro entitlement — your purchase status is stored locally and verified through Apple's StoreKit. Mimic does not maintain its own user account system.
The Community Library
Mimic 1.3 added a library of voices people have chosen to share. Using it is optional, and browsing it tells us nothing about you — no account, no profile, and nothing that identifies you or records which voices you looked at.
If you publish a voice
Publishing is deliberate: you pick a voice you created, choose up to three words that describe how it sounds (and a short note if you want one), and confirm on a separate screen that you have permission to share it. When you do, Mimic sends to our server:
- the voice's 256-number speaker embedding — a mathematical signature, not audio, and not reversible into your recording;
- the name and description you typed;
- the thumbnail image attached to that voice, if it has one.
Your recordings are never sent. There is no way for another user to hear your actual voice sample.
A published voice is public: anyone using Mimic can see its name, description and thumbnail, and can add it to their own app. Please do not put personal information in the name or description, and do not use a photo you would not want to be public.
Moderation and takedown
- Nothing appears automatically. Every submission is reviewed by hand before it becomes visible to anyone. A submission that is not approved is never shown.
- Anyone can report a voice from inside the app, with a reason. Reports reach us immediately.
- Reported voices are hidden automatically once enough reports arrive, without waiting for us to read them, and stay hidden pending review.
- You can remove your voice at any time from the app. It disappears from the library; copies other people already added to their own device remain on their device, as we have no way to reach into it.
- To ask about a published voice, or to report one outside the app, email the address at the bottom of this page.
What the server keeps
The server stores approved voices and, for each, a count of how many people liked and downloaded it. It keeps reports so we can act on them. It records the IP address of publish and report requests for a short period, purely to rate-limit abuse. It also holds the anonymous usage counts described under Anonymous usage counts, which carry no identifier and are not connected to any voice or person. There is no account and no profile, and the like and download counts are plain totals, not records of who did what.
The publisher tag
Publishing to the Community Library is entirely optional, and the library works the way any shared space does: a voice has a creator. When you publish your first voice, Mimic creates a random code and stores it on your device, so the library can tell that several voices came from the same creator.
That is what makes blocking work. Block a creator and every voice of theirs disappears from your library, including whatever they publish next — without the code, blocking could only hide one voice at a time. It also keeps any one person from flooding the review queue.
- It is random — not derived from your device, your Apple Account, your name or anything about you, and not Apple's advertising identifier.
- It is only created if you publish. If you never publish a voice, one is never made.
- We never store the code itself — only a one-way hash of it. Other people see a short fragment of that hash and nothing more.
- It is not used for advertising, analytics or tracking, and it is never shared with anyone else.
- Nobody can look you up by it. It is never shown as a name, and it links to nothing but the voices you chose to publish.
Because the code persists, Mimic's App Store privacy label lists Identifiers alongside User Content. It is used for app functionality and nothing else.
Blocking
You can block a creator from inside the app by pressing and holding any voice in the Community Library. Every voice from that creator disappears from your library, including anything they publish afterwards. Blocking happens entirely on your device — we are not told about it — and you can undo it any time in Settings → Blocked Creators.
Permissions Mimic may request
Mimic asks for permissions only at the moment you opt into a feature that needs them. You can revoke any of them in iOS Settings at any time.
- Microphone — to record your voice in Performer Mode and for cloning.
- Camera and Photo Library — only if you choose to attach a picture to one of your voices. Photos you pick stay on your device unless you publish that voice, in which case the picture becomes its public thumbnail.
- Notifications — when granted, Mimic tells you when a voice you published has been approved and is live in the Community Library, and badges its icon when new voices arrive there. These are local notifications, created on your device. No remote push servers are used, and Mimic sends you nothing about trials, offers or prices.
That is the entire permission surface. Mimic does not request Contacts, Location, Calendar, HealthKit, Speech Recognition, or any other capability.
Anonymous usage counts
Mimic counts which features get used, so we can tell what is working and what to fix. This is the Usage Data entry on Mimic's App Store privacy label. It is not linked to you, and there is no identifier of any kind in it — not the publisher code, not a device id, not an install id, nothing that could be joined back to a person or to a published voice.
When you leave the app, Mimic may send a single summary containing:
- whether this was a first launch, whether the app was opened on a day after the first (counted once a day, so we can tell how many people come back), and whether you are on the free tier, a trial or Pro;
- your iPhone model and the major iOS version — so we know which devices can run the voice cloning;
- counts of feature events, such as how many voice clones were started and how many finished, and — if you subscribe or buy — which of the two products it was, so we can tell a trial starting apart from a purchase. No price, no transaction, no receipt: Apple handles all of that and we never see it.
- which of Mimic's own fourteen built-in voices — the six free effects and the eight designer voices — was used or marked as a favourite, so we can tell which of them earn their place. A voice you cloned, or took from the Community Library, is never named — those names are written by people, and no name a person typed ever leaves your iPhone.
- how far you have got on Mimic's progress ladder — your rank, and the length of your current run of days as a range such as “3–6” rather than an exact number — so we can tell whether people come back. Your points, your medals and the voices behind them stay on your iPhone.
What is never included: your audio, your recordings, the names or descriptions of the voices you make or take, anything you type, and any free text at all. The counts are totals only — there is no record of who did what, because there is nothing in them to identify anyone.
If you turned these counts off, they stay off. Mimic 1.3.5 and earlier offered a switch at Settings → Analytics → Share anonymous usage; if you used it, that choice is still honoured by every later version — the counts stop immediately and permanently, and nothing about how Mimic works changes.
Newer versions no longer show the switch. The counts carry no identifier, so there is nothing in them that can be traced back to you, to your iPhone or to a voice you published — which is why Mimic stopped asking for permission to count rather than offering a control over data that is not about any individual.
What Mimic does not do
- We do not upload your audio. Recordings and voice samples never leave your device.
- We do not use third-party analytics, and nothing you record, say, type or pick is ever measured. See Anonymous usage counts for the small amount we do count.
- We do not include third-party SDKs for advertising, tracking, attribution, crash reporting, or social media.
- We do not require an account. Publishing a voice attaches a random creator code so blocking works — see The publisher tag. It is not used for advertising, analytics or tracking.
- We do not sell, share or rent your data.
- We do not track you across other companies' apps and websites.
On-device machine learning
Mimic's voice cloning uses a Core ML adaptation of the open-source OpenVoice v2 model. Both the speaker-embedding extractor and the voice converter run locally on your iPhone's Neural Engine. Inference never touches the network, and no inference output leaves the device unless you explicitly share it through iOS's share sheet or publish the voice.
The first launch may download additional Apple-provided audio frameworks the OS uses on-device; these are managed by iOS and not by Mimic.
Sharing what you create
When you tap the share button on a converted recording, iOS opens its standard share sheet. The audio file is handed off only to the destination you pick (Messages, Mail, AirDrop, Files, etc.). Mimic itself does not see or relay the content of the share.
Purchases
Mimic Pro is sold through Apple's StoreKit as either a monthly subscription, which starts with a 3-day free trial, or a one-time lifetime purchase. Apple processes the transaction; Mimic receives only the entitlement status and never sees your payment details. To cancel a subscription, use the Subscriptions section of your Apple ID in iOS Settings.
Acceptable use
Mimic is intended for personal creative use — performing, voicing characters, narration and dubbing for your own videos, accessibility, and similar. Using Mimic to impersonate someone without their consent, commit fraud, harass another person, or produce non-consensual sexual content is prohibited by Mimic's terms of use and likely illegal in your jurisdiction. Publishing a voice that is not yours to share is prohibited, and published voices that break these rules are removed.
Children's privacy
Mimic is not directed at children under 13 and we do not knowingly collect personal information from them. The Community Library is moderated before anything becomes visible, and Mimic has no accounts, profiles, comments or messaging, so users cannot contact one another. If you believe a child has published personal information through Mimic, email us and we will remove it.
Changes to this policy
If Mimic changes how it handles data, we will update this page, bump the effective date above, and surface the change inside the app before it takes effect.
Contact
Questions, concerns, or a takedown request? Email gwolny@me.com.