Seshari

Turn any voice message, audio or video into clean text — 100% on your device.

No cloud, no account, nothing collected. Strong in English, Russian and Ukrainian.

Write, and I will tell you when it ships Not in the App Store yet.

One app for Mac, iPhone and iPad — macOS 14 and iOS 17 or later.

Seshari with a transcribed voice message open: the player, its one-line summary, the tags and the transcript with a timestamp on every line.

Private by design

  • Everything runs on-device
  • No account, no sign-in, no ads, no analytics, no tracking
  • Your audio and your transcripts never leave the device

App Store privacy label: Data Not Collected.

The speech model (about 626 MB) is downloaded once, on first launch. After that, everything runs offline.

This page is built the same way

It loads nothing from anyone else — no fonts from a CDN, no analytics, no cookies, no third-party script. Open the network tab and count the hosts: one.

What you get

Made for the voice notes people send you in chats (it opens the .ogg/.opus files messengers export), voice memos, interviews, lectures, meetings and podcasts.

  • Drop in or share any audio or video and get the text back
  • On-device dictation — speak, get text, paste anywhere
  • An automatic title, summary and keywords for every transcript
  • Read your transcript aloud
  • Export as text
  • On the Mac: a global hotkey for dictation, the Services menu, Shortcuts and a URL scheme.
The empty state on the Mac: a drop zone that says «Drop a voice message here», a «choose a file…» link and a Dictate button.
Drop a file, choose one, or dictate.
Searching the library on the Mac for «call»: one note matches, and its transcript is open on the right.
Search across your notes — on the Mac, a global hotkey away.
Seshari on the iPad: a transcribed voice message with its summary, tags and the transcript with timestamps, the sidebar collapsed.
The same app on the iPad.

Strong with Russian and Ukrainian

English, Russian and Ukrainian are first-class here, not an afterthought.

Five on-device models to choose from, downloaded once and kept on the device; the recommended one is about 630 MB, the smallest 73 MB.

The models screen on the iPhone: Tiny, Base, Small, Large v3 turbo (recommended) and Large v3, with their sizes and a Download button each.
Pick a model; change or delete it any time.

Free, and what Pro adds

Free transcribes any file up to 5 minutes, as many as you like, and dictation has no limit at all.

Seshari Pro is a one-time unlock, not a subscription:

  • Files of any length — meetings, lectures, podcasts, interviews
  • Chapters, and a summary for each chapter, on long recordings
  • Several files at once
  • Subtitle export: SRT, VTT and JSON
  • Meaning-based search across your notes
  • On the Mac: watch a folder — new recordings transcribe themselves

$9.99, once.

Pay once. No subscription, no account, no cloud.

A twelve-minute interview on the Mac, split into chapters: Chapter 1 open with its one-line summary and the transcript with timestamps, Chapters 2 and 3 folded, a Transcript / Digest switch above.
Pro: chapters, each with its own summary.
Transcription settings on the Mac — model, language, translation, folder watch — and the Seshari Pro card: unlimited length, chapters and chapter summaries, batch and folder watch, «A one-time purchase. No subscription.»
Pro on the Mac: the card in Settings, and folder watch.

Seshari is not affiliated with or endorsed by any messaging service; it simply reads the standard audio files they export.