Does turning off Speech Recognition delete your existing transcripts?

This question usually comes up after someone has already been using a transcription app for a while, then goes digging through Settings — maybe cleaning up permissions, maybe just curious what is turned on — and finds a Speech Recognition toggle they do not remember granting. Turning it off feels like it should be reversible in the scary direction too: if granting it let the app transcribe, surely revoking it undoes what was transcribed. It does not, and the reason is the same reason a lot of iOS permissions work the way they do.

What the Speech Recognition permission actually is

It lives in Settings → Privacy & Security → Speech Recognition, listed per app, separate from Microphone. Any app that calls into Apple's Speech framework — the same framework behind keyboard dictation — needs this permission granted before that framework will do anything for it. It is a single, system-wide gate: you either allow an app to request speech recognition, or you do not.

That single gate is broader than it sounds. It does not just cover the case where audio goes to Apple's servers. Apple treats turning speech into text as sensitive enough that the permission applies to on-device recognition too, even though nothing leaves the phone on that path. The framework checks authorization status before it will run a request at all, and it does not ask which processing location you are worried about.

It blocks requests, not files

A permission in iOS controls whether an API call is allowed to happen right now. It has no mechanism for reaching backward into files an app already wrote to its own storage — there is no hook that says "and also delete anything this capability previously produced." That would be a strange, destructive thing for a permission toggle to do, and it is not how any of them behave: turning off Photos access does not delete photos already imported into an app, turning off Contacts access does not delete contacts already synced, and turning off Speech Recognition does not delete transcripts already written to disk.

A transcript, once it exists, is just text sitting in the app's own storage on your phone. It was produced by a Speech Recognition request at some point in the past, but it does not stay tethered to that permission afterward — the request finished, the text got saved, and from then on it is an ordinary file like any other, no different from a note you typed by hand.

What actually stops working

• Transcribing a brand-new recording — the request never starts, on-device or otherwise, because the framework refuses it before any audio is looked at.
• Re-transcribing an old recording, if you ever wanted to try again with a different language, or after Apple has shipped a better model — same block, same reason.
• Nothing about playback, editing an existing transcript by hand, exporting it, or searching it. None of those go through the Speech framework, so none of them are affected.

Recording itself is unaffected too, and this is a common point of confusion because both permissions sound related. Recording audio uses the Microphone permission, which is entirely separate from Speech Recognition. You can revoke Speech Recognition completely and keep recording exactly as before — you just will not get a transcript out of anything recorded from that point until you turn it back on.

Turning it back on

There is nothing to restore, because nothing was removed. Flip the toggle back on in Settings, and the next transcription request goes through the framework the same way it always did — on-device attempted first, for whichever languages have a local model, with the usual retry if that pass comes back empty. Transcripts made while the permission was off simply do not exist yet, because they were never requested; everything made before you turned it off was already saved and was never at risk.

How Voice Studio fits into this

Voice Studio stores recordings and transcripts in the app's own storage on the device, not behind an account and not synced through CloudKit, so there is no remote copy for a permission change to interact with either way. If Speech Recognition access is off, the app cannot start a new transcription and will not produce one until it is turned back on — but anything already transcribed stays exactly where it was, readable, editable and exportable the whole time. The permission decides what the app is allowed to do next, not what it is allowed to keep.

Common questions

If I turn off Speech Recognition, do my old transcripts disappear from the app?

No. They were already saved as text before the permission changed, and a permission only controls new requests — it has no way to reach back into files that already exist.

Does this also stop recording audio?

No. Recording uses the Microphone permission, which is separate. You can turn off Speech Recognition and keep recording normally; you just will not get a transcript for anything recorded until you turn it back on.

Does the Speech Recognition permission only block the server fallback, or on-device transcription too?

Both. It is a single permission covering the whole capability. Apple gates on-device recognition behind it as well, so turning it off stops a local-only transcript from starting, not just one that would have gone to a server.

Where do I find this setting?

Settings → Privacy & Security → Speech Recognition. It lists each app that has ever asked for the permission, with its own on/off toggle.