Notarius: Transcribe Offline

Notarius records a conversation, turns speech to text, tells the speakers apart, and rewrites it into a clean, structured note — offline transcription that never leaves your phone.

No account. No upload. No server. Once the models are downloaded, offline transcription works in airplane mode.

WHY IT IS DIFFERENT

Most speech to text apps send your audio to a server. Notarius never does. The speech recognition, the speaker separation and the AI that organizes your note all run on the device in your hand. Your audio, your transcript and your notes never leave it — there is nothing to opt out of, because nothing is collected.

WHAT IT DOES

• Record a conversation with the built-in voice recorder, or import audio you already have
• Offline transcription in 23 languages — audio to text with no connection
• Speaker labels: Notarius works out who spoke when, automatically
• Teach it your voice once, and your own turns get marked as yours
• Rename and merge speakers; the change applies to the whole transcript
• On-device AI turns the raw transcript into a tidy, structured note
• Choose a note style, or write your own prompt
• Tap any line of the transcript to hear that moment of the recording
• Full-text search across every note
• Notes are plain Markdown files you own — export or share any note, or zip up all of them

WHAT PEOPLE USE IT FOR

• Meeting notes and meeting minutes, written up before you have left the room
• Interview transcription, for journalists and researchers
• Lecture notes — record a lecture, get a structured summary
• Dictation and voice memos you would rather keep to yourself
• Supervision and client sessions that have to stay confidential

BUILT FOR CONFIDENTIAL CONVERSATIONS

Therapists, doctors, lawyers, researchers and journalists handle conversations that must not be uploaded to a third party. Notarius was built for exactly that: a voice recorder and note-taker with no cloud behind it. Because nothing is transmitted, no third party ever receives your audio.

HOW IT WORKS

1. Record, or import a file
2. Notarius transcribes it and splits the transcript by speaker
3. Ask it to make notes — the on-device AI reorganizes the transcript into headings, decisions and action points
4. Edit anything; the transcript and the note stay linked

PAY ONCE

Seven days free, then a single purchase. No subscription, no credits, no per-minute fees, free updates. Your notes stay readable whether you buy or not.

BEFORE YOU START

Notarius downloads its AI models on first run. They are large — plan for several gigabytes and use Wi-Fi. Japanese, Korean and Chinese use a second speech model, downloaded the first time you choose one of them.

Turning a transcript into a note runs a language model on the device, which needs a phone with at least 6 GB of memory. Recording, transcription, speaker labels and search work on every supported phone.

After that, the app needs no connection at all. The only network traffic Notarius makes is that model download and the Play Store purchase itself.