The iPhone and iPad app
The app does everything the web editor does, on the phone: transcribe, style, fix words, retime, and save a captioned video. There is no account and no server, and nothing is uploaded. It needs iOS 18 or later and works on iPhone and iPad.
Make captions
Section titled “Make captions”- Add a video from your photos. You see it with its length and size, and can rename it, before it is imported.
- Transcribe. The sheet asks, as the web app does, for the speech model and then the language. The first use of a model downloads it, and a card shows the progress.
- Edit. The video takes the screen with a compact timeline under it. Tap to play or pause. While paused, drag the caption to move it (it snaps to the centre lines, with a small haptic tick), pinch to change the size, and double-tap a word to fix it. Drag the edges of a caption on the timeline to retime it, pinch the timeline to zoom, and scrub by dragging it.
- Style. Open the style controls: the same thirteen looks and the same settings as the web.
- Save. Choose the size and frame rate. Videos are saved as H.264, or 10-bit HEVC when the video is HDR, and HDR videos stay HDR.
Your work is saved by itself, after a quiet moment and whenever you leave the app. There is no Save button. Undo and redo are in the toolbar.
Speech models on the phone
Section titled “Speech models on the phone”| Model | Download | Notes |
|---|---|---|
| Whisper Small | 217 MB | The smallest, fastest to try. |
| Parakeet v3 | 480 MB | Fast, for its 25 European languages. |
| Whisper Large v3 Turbo | 627 MB | The default, and the best all-round accuracy. |
| Whisper Large v3 | 948 MB | The most accurate Whisper. |
| Apple Speech | Built into iOS 26 and later | Needs nothing to download. |
Models are downloaded once and run on the phone. Settings lists them with their size on disk, and you can delete the ones you do not use. The language list follows the model, as it does on the web.
While a long video is transcribed, the app tells you to keep it open until it finishes. If the speech model skips a loud stretch, a second pass re-reads it and keeps only the words two readings agree on.
Use your own server instead
Section titled “Use your own server instead”If you run an OpenCaptions server with a graphics card, the phone can send the audio there for a faster transcription. The video itself never leaves the phone, only its audio, and only if you set this up.
- In the web app, open Account and create an API key in the API keys section. It comes with a QR code and a link.
- On the phone, scan the QR code with the Camera and tap the banner that appears. Or open Settings, find Where to transcribe, and paste the link, or enter the address and the key.
- Choose Remote server there. The app asks before it accepts a server from a link.
The app says whenever audio is about to be sent to a server, and names its host.
Free and Pro
Section titled “Free and Pro”The free app has everything you need to caption a video and save it up to 1080p at 30 fps, with a small mark in the corner. Pro is a single purchase, with no subscription: it removes the mark and adds 4K, 60 fps, HDR, every caption style and the most accurate speech model. The code is the same either way, and a build you make yourself has everything unlocked.
Privacy
Section titled “Privacy”The app has no analytics and no tracking. Your videos stay on the device. The only network requests are model downloads, Google Fonts for a font that is not bundled, and the server you connect yourself. See the privacy policy.