Support
Seshari — voice messages to text, entirely on your device.
Something broken, or an idea? Write to support@seshari.app — it reaches me directly, and a real person answers. Telling me your macOS or iOS version and what you were doing makes it much faster to fix.
Common questions
- Is it in the App Store?
- Not yet. Write to support@seshari.app and you will hear from me when it ships — there is no list to sign up to, because there is no server to keep one.
- Which model should I pick?
- The recommended one, Large v3 turbo (about 630 MB): the best all-round, strong in English and solid in Russian and Ukrainian. Tiny (73 MB) is the fastest and roughest, Base (~150 MB) and Small (~500 MB) sit between, and Large v3 (~3 GB) is the most thorough on English with no gain over turbo on Russian. All run on the device; change or delete them any time in Models.
- Why does it download something on first run?
- The speech-recognition model. It is downloaded once, then transcription runs fully offline. You can pick a smaller, faster model instead of the recommended one, or skip the download and do it later.
- The first transcription took much longer than the ones after it.
- That is expected once, per device: the system compiles the model for the Neural Engine the first time it is used. Later runs reuse the compiled version and are dramatically faster.
- Does my audio leave the device?
- No. There is no upload path in the app. See the privacy policy.
- Which languages does it handle?
- Whisper's full language set, with particular care taken over Russian and Ukrainian. The interface itself is English, Russian and Ukrainian, and a fresh install opens in your system's language when it is one of those.
- Which files can it read?
- Telegram voice messages (.ogg, .oga, .opus), plus .m4a, .mp3, .wav, .aac, .flac, .amr, .aiff, .aif, .caf, and the video containers .mp4, .mov, .m4v and .3gp — the audio track is used. WebM and Matroska are not supported. A shared file with no extension is sniffed, so Telegram's unnamed voice notes still work.
- I bought Pro and it is not showing.
- Use Restore Purchases — on the upgrade screen, and in Settings. A purchase is tied to your Apple Account, so make sure you are signed in with the one you bought it with.
- What is free and what is paid?
- Free covers unlimited notes up to five minutes each, dictation of any length, automatic titles, keywords and summaries, and TXT/Markdown export. Pro is a one-time purchase that removes the length limit and adds chapters for long recordings, on-device generative summaries, batch and folder processing, and SRT/VTT/JSON export.
- Where are my transcripts stored?
- In the app's own storage on your device, as plain files. Deleting a note removes its text and its audio copy; deleting the app removes everything.