Offline voice changer · iPhone
Hold to speak.
Release to transform.
Your delivery, your pauses, your accent — all kept. Only the voice changes. Six character voices, every millisecond of it computed on your own device. No upload, no connection, no account.
In review. This turns into a download link the day it ships.
iPhone · iOS 26+
Six character voices
Release to transform & play
No settings. One button.
Nothing to type, no model to pick, no upload to wait on. From press to playback there is no third step.
- 01
Hold
Press the button to record, up to thirty seconds. The ring counts down; slide up to cancel.
- 02
Release
Conversion starts the moment you let go. It never transcribes your speech, so it can never mishear it.
- 03
Hear it
Plays back automatically with a real amplitude waveform. Replay it, or export a WAV and send it on.
Six character voices
Not a filter. Someone else says your sentence — same timing, same feeling, different throat.
Hero
Bright and certain. Built for declarations.
Knight
Steady and thick, with a faint metallic weight.
Overlord
Low and looming. Every sentence sounds like an order.
Pikeman
Rough and plain, tired from the march.
Robot
Cold, flat, regular — the mechanised take on a voice.
Townsperson
Ordinary and loose, like someone chatting on a corner.
Record three to five seconds to build a voice of your own. It stays on the device too.
Your voice never left the device
That is not a policy promise. It is a fact about the code: the app has no networking in it.
- Zero requests
- There is no networking logic in the source and no network entitlement. Turn on airplane mode; nothing changes.
- Zero accounts
- No sign-up, no login, no email. Open it and use it.
- Zero collection
- No analytics, no tracking SDK, no crash reporting. We cannot see anything you do, recordings included.
- Processing stays local
- Audio is processed directly on your device and is never sent to us or a third-party server.
Why it can be this fast
The usual approach transcribes you, then reads it back — slow, and wrong whenever it mishears. Mimic skips text entirely and converts the voice directly.
- ~360ms
- to convert ten seconds of speech
- 0.46s
- cold start to ready
- 0
- voice data sent to the cloud
Questions
Does it need an internet connection?+
No. After the initial App Store download it is fully offline — airplane mode makes no difference.
How much space does it take?+
Around 900 MB of voice models, installed with the app. That is the cost of keeping the computation on your device.
Can it imitate a specific real person?+
The built-in voices are fictional characters. You can record your own voice to make a voice, but impersonating other people is not something we support or encourage.
How long can a recording be?+
Up to thirty seconds per take. At the limit, recording stops and conversion begins automatically.
Which devices are supported?+
Mimic requires an iPhone running iOS 26 or later. It uses the Neural Engine and the GPU.
Can I use the converted audio elsewhere?+
Yes — export it as a WAV and share it.
Hold to speak. Release to transform.
One button. Six characters. No connection.