Skip to main content

Legal & policies

How voice transcription works

Last updated: 15 October 2026.

Transcription happens on your own device

MySymplyApp turns your voice notes into text locally, on the device you are using. A speech-recognition model (Whisper) runs inside your browser. The saved recording is not sent to OpenAI, to the Lovable AI Gateway, or to any other outside speech-to-text service in order to produce the standard transcript.

Once your device has produced the text, the transcript is saved back into your MySymplyApp account (hosted in the European Union) alongside the recording, so you can read it, correct it and use it later.

Saving your private recording to EU account storage is separate from speech-to-text processing. Local transcription does not mean the recording is never uploaded: the saved audio can be replayed from your account, but is not sent to an outside transcription service.

You choose whether transcription runs automatically

  • When it is on: a saved voice note is transcribed on your device after it is saved, while the app is open and your device can run the model. Closing or backgrounding the app may interrupt it; this is not a background service.
  • When it is off: nothing is transcribed automatically. You can still transcribe an individual recording later by tapping Transcribe now on the entry.
  • Turning it off later: recordings that were already transcribed keep their transcripts. Delete the entry to remove both the recording and its transcript from MySymplyApp.

Reviewing and correcting the text

Automatic transcripts are drafts. The original automatic transcript is kept separately from any corrections you make, so your own wording is never silently overwritten and the original is never lost. Until you have reviewed a transcript, its text is not included in approved or shared reports — only the entry itself (date, duration, mood, your notes) appears.

The optional “Check for possible mistakes” tool also runs entirely on your device. It makes no network call and never changes your words without you accepting a suggestion.

If your device cannot finish the transcription

Some older devices or browsers cannot run the speech model. If transcription fails or is interrupted, your recording is still saved and the entry is marked so you can retry on that device or a different one, or type the text yourself. There is no cloud fallback: MySymplyApp will not send your audio to an outside transcription service instead.

What this means for your data

  • Voice audio for standard transcription does not leave your device for speech-to-text processing, and is not transferred outside the EEA for that purpose.
  • The recording itself, if you keep it, is stored in MySymplyApp's encrypted storage in the European Union under your control.
  • The speech model files are downloaded to your browser from a public model host the first time you transcribe. That download sends none of your recording, transcript, health information or account details — it only requests the model files. As with any file downloaded from the internet, your device's connection details (such as your IP address and browser type) are visible to that host under its own terms.
  • We do not claim this arrangement carries “zero privacy risk”, and it has not been independently certified or legally approved.

Other ways to limit transcription

  • Turn off Automatically transcribe my voice notes in your account settings. When signed in, this lives at Account & Privacy.
  • Delete an existing recording — this deletes the audio file and its transcript from MySymplyApp.
  • Withdraw consent — MySymplyApp stops processing new health data, including transcription, while your existing data remains until you request deletion.

Your retention controls

You control how long the original audio is kept: delete after transcript approval, keep for 7, 30 or 90 days, or keep until you delete it. Deleting audio does not delete the reviewed transcript.

A separate feature: document analysis

MySymplyApp Scan (reading letters and documents) is a different, separately opt-in feature and does use an outside AI provider. See how document analysis works and the sub-processors page.