transcribe on your phone
Words on the device, no desktop required.
yapyap go can transcribe a recording on the phone itself. No desktop, no network — it runs on the device, and the transcript never leaves it. Useful when you travel, or when the desktop is off and you need the words now.
you already have one
Out of the box the phone uses its own speech recognition — Apple's on iOS, the on-device recogniser on Android. Nothing to download, nothing to choose, and it still runs offline: yapyap always asks for the on-device half of those APIs, never the half that would send your audio to a server.
That is the default, so you can skip this whole page and just record.
or download a model
A downloaded model hears more accurately, at the cost of megabytes and time per recording. Open Settings → Transcription on the phone and pick one:
| engine | size | good for |
|---|---|---|
| your phone's own | no download | the default — offline, instant, no setup |
| tiny | 32 MB | fastest of the downloads, rough around the edges |
| base | 60 MB | the sweet spot for voice notes |
| small | 190 MB | most accurate, slowest download |
| parakeet | 356 MB | the desktop's engine — english + european languages only |
The three whisper models understand many languages and detect the language on their own. The onboarding wizard offers the same picker — if you chose there, you are done.
Some phones cannot recognise speech offline — an Android without the on-device recogniser installed, or an iOS whose language was never downloaded. yapyap notices, says so on that row in the picker, and falls back to downloading base rather than putting your audio on someone's server.
transcribe a recording
- Open a recording's options and tap transcript.
- Tap transcribe. A progress percentage counts up — you can stop it at any time.
- Read, share transcript, or transcribe again with a different model.
Transcription runs one recording at a time, with the app open.
how this relates to the desktop
The phone transcript is a convenience copy and it stays on the phone. When the recording syncs to a paired desktop, the desktop transcribes the audio again with its bigger models — that transcript is the one your lenses and search use.