Apograph

Records, transcribes and tells apart who said what — entirely on your iPhone, in 99 languages. No account, no cloud, nothing uploaded. Pro is a one-time purchase.

Apograph turns speech into text on your iPhone — and only there. Recordings, transcripts and speaker labels never leave the device. No account, no server, no analytics.

RECORD OR IMPORT • Tap record, ask Siri, or use the Control Center tile or the Action button. • Import audio or video — a Voice Memo, a meeting, a saved lecture — from Files, Photos or the share sheet. • Auto-stop ends a long recording at the minute you set — and before the iPhone runs out of space, the capture stops itself, saved and transcribed. • A Live Activity shows the running timer — and, with Auto-stop, when the recording will end — with a Stop control on the Lock Screen and in the Dynamic Island.

TRANSCRIBED ON THE DEVICE • OpenAI Whisper models run locally on the iPhone's GPU. Works in Airplane Mode. • 99 languages, detected automatically — including languages the iPhone's built-in transcription doesn't support. • Three model tiers: Fast ships with the app; Balanced and Best download once and stay on the device. • Stable decoding — no random temperature drift between runs.

A TRANSCRIPT YOU CAN WORK WITH • Timed sentences: tap any sentence to hear just that passage. • Edit the transcript in place — fix a word, a name, a whole sentence. • A spell checker that learns your names and terms — remembered words are fed back to the model. • Retranscribe a sentence — or the whole recording — with the Best model, and compare before replacing anything. • Export as text, Markdown or PDF. Copy or share with one tap.

YOUR FILES, IN FILES Every recording is a plain folder in the Files app — audio, metadata and transcript.txt — with readable date names. Store them in any folder you choose. They are yours.

APOGRAPH PRO — ONE PURCHASE, NO SUBSCRIPTION • Best model: the highest transcription accuracy. • Speaker separation (diarization): who said what — the voices within a recording are told apart on the device, labeled for you to rename, and carried into every export. It never identifies whose voice it is. • Subtitle export: SRT and WebVTT with real timings. • Unlimited Best re-transcriptions — whole recordings or single sentences. Free includes unlimited recordings, the Fast and Balanced models, every language, text, Markdown and PDF export, and the Best model on five recordings — full or per-section.

PRIVACY The app contains no analytics, advertising or tracking SDKs. The only network traffic is fetching the model catalog, downloading a model you chose, and the App Store handling your purchase. Full policy: getapograph.io/privacy

Requires iOS 18.6 or later. The Best model needs more memory than some iPhones have; the model picker tells you which tiers fit before you download anything.