SpeechDone

Said.
Done.

A message, a note or a whole meeting: SpeechDone turns your voice into text you can send, keep or share. On iPhone, iPad, Mac and Apple Watch.

One subscription gets you all three: talking instead of typing, meeting notes and memos.

You confirm by email. We write once at launch, then delete your address. Privacy

Our server is in Germany, and your speech is processed only in the EU.

Speak

Think out loud. Send it like you wrote it.

Say “Tuesday, uh, no, Wednesday,” and your text says Wednesday. It works with the SpeechDone keyboard in every app, and on the Mac you hold the fn key in the bottom left corner of the keyboard.

Nobody likes thumb-typing a real reply. So it waits until you’re back at your desk.

A voice memo just hands the work to the person on the other end.

With SpeechDone you say the reply now, and it still reads like you typed it.

What you said: Quick heads-up on the proposal, um, I can get it to you by Tuesday, uh, no, wait, Wednesday. The numbers, like, those’ll come first thing tomorrow.. Pasted: Quick heads-up on the proposal: I can get it to you by Wednesday. The numbers will come first thing tomorrow.

Before and after

You said this. The text box got this.

What you said

Pasted

A real recording from the app. We didn’t touch the output.

Why SpeechDone

You talk, and SpeechDone writes it down and cleans it up.

  1. Your text is there the moment you stop.

    You see your words while you talk. When you tap stop, the finished text is usually already in the field. In our test on October 13, 2026, at least half of all recordings needed no wait at all.

    0 msFor at least half of all recordings, the text was already in the field the moment the speaker tapped stop (measured October 13, 2026).
    You talk, the words appearStop: the text is there

  2. Your words show up while you talk, and the final text is more precise.

    While you talk, your words already appear on screen. At the same time, Mistral, a speech recognition company in Paris, listens to your recording again, piece by piece, and gets it more precise. Usually that more precise text is ready when you stop, and it goes into the field. If not, you get the text you saw while talking. No other provider recognizes your speech.

    While you talk, Mistral writes along and you see the words.
    At the same time, Mistral listens to each piece of your recording again, more carefully.

    When you stop, the more precise text goes into the field. If it isn’t ready yet, you get the text you already saw.

  3. You can talk on iPhone, iPad and Mac.

    On iPhone and iPad you talk through the SpeechDone keyboard, in any app where you would normally type. On the Mac you hold the fn key in the bottom left corner, talk and let go, and the text appears where you are typing. Later you find what you said as text in your History, saved in your own iCloud.

    A keyboard in every app, and right at the cursor on the Mac.

1 · Talk freely

Reply to your boss between meetings. Leave the ums in the hallway.

Your boss needs an answer by three, and you’re already walking to the next room.

So you talk it through, false starts and all. SpeechDone starts cleaning up in every pause, and when you tap stop, it drops into your draft.

Punctuation in. Ums out.

2 · Casual in chats, polite in email

Text the family chat while you cook, as casually as you talk.

One hand’s on the spoon, the other’s holding your phone, and the group chat wants to know who’s bringing dessert Sunday.

In the group chat, SpeechDone keeps it loose, the way you’d actually say it to your friends. An hour later the same voice comes out buttoned-up in an email to a client, because SpeechDone follows the app you’re typing in.

When you ask an AI app a question out loud, every word stays. SpeechDone only adds the punctuation.

Add names and jargon to your dictionary once, and SpeechDone keeps the spelling right.

3 · Recording kept

Say that long request to your AI app on the subway. If the signal drops, your recording’s still there.

You think out loud for two minutes, then the train goes underground.

The moment you tap stop, the whole recording is saved on your iPhone. If it doesn’t get through, SpeechDone sends it again the next time you open the app.

In your History, your original stays next to the cleaned-up version.

Meetingscoming in an update (version 1.1)

SpeechDone takes the notes. You listen.

After the meeting, SpeechDone has your notes, split by speaker, plus a short summary, the decisions and the tasks. No extra participant joins the call to take notes, and nobody has to share a screen.

  • On the Mac, SpeechDone records your microphone and the sound from Teams, Zoom or Google Meet.
  • In the room: put your iPhone or iPad on the table.
  • On your wrist: Apple Watch records the whole meeting.
  • Before every recording, you confirm that everyone knows.

With Pro, 10 hours every 30 days. Free and Starter include one meeting to try, up to 30 minutes.

Sketch of a meeting transcript

Weekly sync

On the Mac · sound from Teams

Summary

The campaign starts on Wednesday. The photos will be picked by then.

Decisions

Start on Wednesday.

Tasks

Pick the photosSpeaker 2

Transcript

Speaker 1Shall we start on Wednesday?

Speaker 2Yes, I’ll have the photos picked by then.

Inbox

  • Get the photos printed for Grandma.

    From Apple Watch

  • Order the cake for Sunday from the bakery.

    From iPhone

Memos

Say the thought out loud before it slips away.

One tap is enough, on Apple Watch too. The memo then waits in your Inbox on iPhone and Mac.

Where your voice goes

Our server is in Germany, and your speech is recognized in the EU.

European Union

  • Germany

    Our server, hosted by Hetzner. Your recording isn’t stored, it’s only processed while it’s being recognized.

  • France

    Mistral AI is a company in Paris that recognizes speech and edits text. It turns your recording into the text that goes into the field and cleans it up. When you rewrite, translate or ask a question by voice, Mistral does that work too.

  • EU servers

    Mistral also writes the text you see while you are still talking, on servers in the EU.

  • Your writing style

    So SpeechDone writes the way you do, the text of your recent recordings stays on our server for up to 30 days. You can turn that off.

USA

No server in the US recognizes or cleans up your speech, not even when a provider in Europe is down. No setting changes that. You can still use SpeechDone wherever you are, including in the US.

Mistral does not use your recordings for training: it is switched off in our account.

The connection between the app and our server runs through Cloudflare, a US company (network and protection against attacks).

Read the privacy policyPrivacy contract for business customers, and the list of our service providers

For your profession

Between appointments: talk instead of typing.

In a medical practice, a law firm or a tax office, it matters where data goes. Our server is in Germany, your speech is recognized on servers in the EU, and you sign the data processing agreement that the GDPR requires with us by email.

With one subscription you can talk on iPhone, iPad and Mac.

iPhone

Its own keyboard in every app. The app itself for longer writing.

  • After your first recording, the mic stays ready, but it only records between start and stop. It stays ready until you turn it off, or for 5, 15 or 60 minutes if you choose.
  • Cleanup starts while you pause
  • Requires iOS 18 or later
iPad with the SpeechDone keyboard in a family chat, recording in progress

iPad

The same keyboard works when two apps are open side by side.

  • The app itself in the iPhone layout
  • The same History through iCloud
  • Requires iPadOS 18 or later

Mac

Hold the fn key in the bottom left corner of the keyboard, talk and let go. The text appears where you are typing.

  • Sits in the menu bar at the top, not among your open apps
  • Rewrite a selection by voice
  • Requires macOS 15 or later

How the keyboard works on iPhone

Apple doesn’t let keyboards use the microphone. So for your first recording, SpeechDone opens for a moment, then your text goes back into the field you started in.

After that, the mic stays ready, so you start your next recordings right from the keyboard. While it’s ready, the top of your iPhone shows it, and one tap turns it off. Only what you say between start and stop is saved or sent.

1 You tap the mic in the keyboard.
2 SpeechDone opens for a moment and records.
3 The text is in the field, and the mic stays ready.

The text is in the field, and the mic stays ready.

What else SpeechDone does

You can translate, ask questions and rewrite the text in a field, all by voice. Phrases you use often are saved as Snippets and inserted with a short code. You can also start SpeechDone with Siri or right from your Home Screen.

Translate, Ask and Rewrite are part of Pro.

Pricing

Start free. No limit stops you mid-thought.

Every plan records, writes and pastes your text, even after you have used your weekly words. From then on, your text appears exactly as you said it, without cleanup, until your new week starts.

Free

$0

1,500 words a week, about 200 a day

You choose whether SpeechDone cleans up, keeps every word, or matches the tone of the app you are writing in.

Starter

$2.99 a month

or $24.99 a year

6,000 words a week, about 850 a day

You choose whether SpeechDone cleans up, keeps every word, or matches the tone of the app you are writing in.

ProRecommended

$4.99 a month

or $49.99 a year, that’s about $4 a month

Join the waitlist

No weekly limit

You choose from all six styles and can rewrite, translate and ask questions by voice.

Speak more than 150,000 words in a month and cleanup slows down a little.

Pro

Recommended

$4.99 a month

or $49.99 a year, that’s about $4 a month

Join the waitlist

No weekly limit

You choose from all six styles and can rewrite, translate and ask questions by voice.

Speak more than 150,000 words in a month and cleanup slows down a little.

Starter

6,000 words a week, about 850 a day

$2.99

a month

Free

1,500 words a week, about 200 a day

$0

Every plan

You get the SpeechDone keyboard, your History in iCloud with the original next to the cleaned-up version, a Dictionary for names and terms, and Snippets: saved phrases you insert with a short code.

How the weekly limit works

At 80%, SpeechDone gives you a heads-up. Hit the limit and you keep talking, just without the cleanup. When your new week starts, the cleanup’s back.

0 80%: heads-up 100%: keep going without cleanup

Subscriptions run through the App Store. Manage or cancel them in your iPhone Settings. Prices in US dollars, the App Store shows your local price.

Frequently asked questions

Where are my recordings processed?

Our server is hosted by Hetzner in Germany. Recognition runs only on EU servers from Mistral AI (Paris). SpeechDone never falls back to US servers, not even when something is down.

Does SpeechDone store my recordings?

Our server does not store your recording. It holds it only while your speech is being recognized and keeps no copy. You then find the finished text in the app’s History. It is stored in your own iCloud, your storage with Apple, not with us.

Does SpeechDone read my screen?

No. To get the tone right, SpeechDone only needs to know which app you are writing in. So the new text fits in, it reads the words right before and after the spot where you are typing, and that stays on your iPhone. Only when you ask SpeechDone to rewrite a text or answer a question about it does it send that text along: what is in the field or what you have selected. It goes to our server in Germany and to Mistral, our speech recognition provider in Paris.

Does SpeechDone work in every app?

Almost always, through its own 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. Apple doesn’t give keyboards the microphone, so SpeechDone opens briefly for your first recording. After that you speak right from the keyboard. On the Mac, hold fn and speak, and the text appears wherever your cursor is.

What happens when I hit my weekly limit?

You keep talking. SpeechDone writes without the cleanup until your new week starts.

Can my practice or law firm sign a data processing agreement with you?

Yes. You sign two documents with us by email. The first is the data processing agreement that the GDPR, the EU privacy law, requires. The second is a confidentiality promise that German law requires when doctors, lawyers or tax advisors share client data with a service provider. The list of service providers we use is public.

When will SpeechDone be in the App Store?

Soon. Join the waitlist and we’ll email you once at launch.

All questions and answers

One subscription gets you all three: talking instead of typing, meeting notes and memos.

For iPhone, iPad, Mac and Apple Watch. Join the list and we’ll email you once at launch.

You confirm by email. We write once at launch, then delete your address. Privacy

Said. Done.