Voice Studio

Does an app need to be in your language to transcribe it on iPhone?

No. It feels like it should be true — if the app has never been translated into a language, why would it understand that language at all? — but interface language and transcription language are two separate settings on iOS, decided by two entirely different things, and neither one limits the other.

It is a reasonable assumption to carry over from other software: an app that only shows its menus in English probably only "speaks" English, the way a person who has not learned a second language cannot suddenly understand one. Speech recognition on iPhone does not work that way, and the reason is worth understanding, because it explains why a small app can plausibly transcribe dozens of languages while having translated its own buttons into only a handful of them.

Two lists, built by two different people

An app’s interface language list is bounded by translation work. Every screen, every button label, every piece of help text has to be turned into each language by someone, checked, and shipped — real, ongoing labor that scales with the number of languages an app’s makers choose to support. It is entirely reasonable for a small team to cover a modest, deliberately chosen set of languages this way and stop there.

An app’s transcription language list is bounded by something else entirely: which speech recognition locales Apple’s Speech framework covers. That list is built and maintained by Apple, ships as part of iOS, and is available to any app that calls it — the app does not train a model, does not localize a recognizer, and does not do per-language engineering to add a new one to its list. It requests a locale, and either iOS has a recognizer for that locale or it does not.

Because one list is hand-built translation work and the other is a shared system capability, there is no reason the two would ever be the same size, and in practice they usually are not.

Why the transcription list is so often the longer one

Translating an interface into a new language is a project: someone has to write or review every string, and a mistranslated menu is a small but real embarrassment that gets caught and fixed. Wiring up a new transcription locale is comparatively cheap for an app to offer, because the actual speech model — the expensive part, the part that took Apple enormous training effort — already exists inside iOS. The app is not the one doing the hard work of understanding that language; it is asking the phone to.

That asymmetry is exactly why it is common to see an app translated into a dozen or so languages while listing three or four times that many transcription options. It is not a marketing exaggeration or an inconsistency — it reflects the fact that displaying a menu and running a speech recognizer are genuinely different kinds of work, done by different parties.

How the two settings actually behave

An app’s own interface language generally follows how your iPhone is set up to show apps in general — the device’s system language, or a per-app language override where iOS offers one — and that is a display choice, nothing more. It changes what the buttons say. It does not touch what the app is capable of doing.

Which language a given recording gets transcribed in is a separate choice, usually made from within the app itself, and it can be any locale the app’s transcription list covers — regardless of what language its own screen happens to be showing at that moment. Set your phone to a language the app has never translated a single word into, and the app will typically fall back to displaying itself in one it has, while the transcription setting sitting a tab or two away is completely unaffected by that fallback.

What this means in practice

The practical habit worth building is to stop reading an app’s interface language as a proxy for what it can transcribe. If a language matters to you, check the app’s actual transcription language list rather than assuming from the menus on screen.

How Voice Studio handles it

Voice Studio’s own interface has been translated by hand into 19 languages. Its transcription setting is separate and covers 38 recognition locales — Apple’s list, not ours, since transcription runs on Apple’s Speech framework rather than anything Voice Studio built itself. The two are independent: you can run the app in any of its 19 interface languages while transcribing in any of the 38 supported locales, in whatever combination actually matches your life, and switching one never moves the other.

Common questions

If my iPhone is set to a language the app has not translated its menus into, will transcription still work correctly?

Yes, as long as that language is on the app’s transcription list. The app’s displayed interface and the language it transcribes are set independently — an interface fallback does not change what a transcription request asks for.

Why does an app support fewer interface languages than transcription languages?

Translating an interface is manual work an app’s makers do once per language they choose to support. Transcription capability comes from Apple’s shared Speech framework, which already exists inside iOS — an app is requesting a locale from it, not building a model for it.

Can I change which language a recording gets transcribed in without changing the app’s display language?

Yes. They are separate settings. Changing one does not move the other.

Does a language’s position on the transcription list tell me whether it runs on-device?

No — that is a third, separate question, decided by whether Apple has shipped a local model for that specific locale and whether it has been downloaded to your phone, not by the interface or the transcription language settings.

Try it in Voice Studio

Voice Studio records, transcribes on your iPhone, and files each note by time and place — so the thought you had in the car is still findable next month.

Download on the App Store

Free to download · iPhone and iPad · iOS 16.4 or later