SpeechDone: Questions and Answers

SpeechDone is a dictation app for iPhone. You talk, and it cleans up already in your speech pauses. After you stop, there’s a text in the field you can send.

You see the text while you speak. After you stop, the finished text is in the field right away, a median of 0 ms after stop (measured October 13, 2026).

SpeechDone cleans them up. Say “Tuesday, uh, no, Wednesday” and your message says Wednesday. Your original stays in your history next to the cleaned-up version.

Two recognizers check each other. Deepgram transcribes live, Mistral (Voxtral) recognizes the recording again in sections, and that text is inserted. If Mistral doesn’t answer in time, the text from Deepgram is used.

Almost all of them: SpeechDone has its own dictation keyboard. The exceptions are password and phone number fields, where iOS always shows Apple’s keyboard, and apps that don’t allow third-party keyboards. It also comes with the app itself, a widget, a Live Activity and the share sheet.

Apple doesn’t let keyboards use the microphone. So SpeechDone records briefly in the app, then your text goes back into the field you started in. After that, the mic stays open, 15 minutes by default.

Yes, in the keyboard. You translate, ask questions and rewrite selected text by voice. These are part of Pro.

Add them to your dictionary once and SpeechDone pays attention to the spelling. Signed in, your dictionary and snippets are the same on all your devices.

Not on our server. It’s only processed in memory during recognition. Your history lives in your own iCloud.

Our server is hosted at Hetzner in Germany. Recognition runs only on EU servers: Deepgram (Deepgram, Inc., USA, processing on EU servers) and Mistral AI (Paris).

No. Recognition and text processing always run in the EU, and there is no switch to the US. If a service is down, the app tries again later.

No. Deepgram and Mistral don’t use your recordings or texts for training: we opt out with Deepgram on every request, and with Mistral it is switched off in our account. To detect abuse, Mistral keeps requests for at most 30 days; details are in the privacy policy.

Yes. You sign the DPA under Art. 28 GDPR with us in text form. There’s also a confidentiality undertaking under § 203 of the German Criminal Code and a public list of subprocessors.

No. To set the tone, it only needs to know which app you’re writing in. In Edit mode it reads the text in the current field, because that’s the text it’s meant to rewrite.

Free costs nothing and includes 1,500 words a week. Starter is $2.99 a month or $24.99 a year (6,000 words a week). Pro is $4.99 a month or $49.99 a year, with no weekly limit.

Your dictation never stops at the limit. At 80%, SpeechDone gives you a heads-up. Past it, you keep dictating, just without cleanup, until your new week starts.

Subscriptions run through the App Store. You manage and cancel them in your iPhone settings.

Not yet. Join the waitlist and we’ll email you once when SpeechDone is in the App Store.

You get a confirmation email and one email at launch. After that, your address is deleted.

iPhone and iPad. On iPad it uses the iPhone layout. It requires iOS 18 or later.