On-device speech-to-text

Voice to text that never phones home.

Scrivo turns speech into text entirely on your device — the recording, the transcription, the search, even the AI that titles your notes. No cloud, no account. Airplane mode isn’t a limitation; it’s the proof.

0servers touching your voice
0accounts to create
0audio files uploaded
1owner of your words — you

The whole point

On your phone. Full stop.

Most dictation apps stream your voice to a data center and mail the text back. Scrivo does the entire job in your hand — the models, the database, the search index. There’s nothing to intercept, nothing to breach, nothing to quietly retrain on.

Data ledgerWhere it goes
Microphone audioStays on device
Imported recordingsStay on device
Transcripts & historyStay on device
Search queriesStay on device
AI titles, tags, action itemsMade on device
Your voice, sent to a serverNever
9:41

Works at 30,000 feet, in a basement, on a mountain. Flip it.

Data leaves only where you point it.

Want transcripts in your own systems? Add webhook endpoints and Scrivo delivers finished notes straight from the phone to your URLs — with retries and per-recording status. Off by default. Your endpoints, your rules.

This website keeps the same promise: no cookies, no analytics, no third-party requests. One origin, static files. View source.

What it does

A recorder that writes everything down.

Scrivo is a complete voice notebook — not a thin client for somebody else’s API.

Live transcription

Watch words appear as you speak. Apple Speech on iOS, ML Kit on Android — both running entirely offline.

Import & share audio

Share a voice note straight from WhatsApp or Telegram, or import any audio file. Scrivo detects the language and transcribes it in the background while you do something else.

Dictate in 30+ languages

Transcription runs on-device in more than thirty languages — and the local Gemma model can translate a finished note, offline, like everything else.

History & search

Every note lands in a local, searchable history with the original audio attached. Find the thing you said in March, in milliseconds.

On-device AI

A local Gemma model writes titles, suggests tags, and spots action items — on your phone’s silicon, not a server farm.

Straight to calendar

Spotted times and dates become calendar events in one tap. No calendar permission needed, ever.

For the tinkerers

Webhooks out.

When a recording finishes, Scrivo can POST the transcript to endpoints you configure — your server, n8n, Zapier, home automation. Straight from the phone to your URL: no middleman, no relay service, with retries and per-recording delivery status.

POST · JSONexponential backoffoff by default
scrivo · delivery log

transcription.completedrec_7f3a · 142 words

POSThttps://n8n.lab.dev/scrivo200 · 84 ms

POSThttps://notes.local/inbox200 · 121 ms

POSThttps://home.io/hooktimeout → retry ×1 → 200

{"event": "transcription.completed","session": {"title": "Morning standup","language": "en-US","wordCount": 142,"transcript": "…stays between you and your endpoint…"}}

Voice to text that never leaves your phone.

No waitlist, no email signup — collecting your address before shipping a privacy app would be a strange way to start. The links light up here the day the stores do.