A teleprompter helps you read.
SpeechMaestro helps you speak.

A teleprompter shows every word and is built for people reading a script to a camera. SpeechMaestro is speaker's notes for people addressing a live audience: the app listens, finds your place in the text and scrolls along while you speak freely. It deliberately shows less — it's meant to lift your eyes off the device, not fix them to it.

For people who speak in front of people for a living: board members and managing directors, ministers and priests, politicians, teachers and lecturers, wedding officiants, association chairs. They know their material and don't need a script to read from — they need orientation: where am I, what's coming next, how am I doing on time.

Probably not. The app earns its place where something is at stake: when names have to land exactly, when figures need to be right, when there's a time slot to hold. For two minutes at the table, a card will do.

It isn't a flashcard trainer; it's orientation for the delivery itself. Indirectly it does help: every run is a full run-through, and marking your accent words forces the decision about what genuinely has to land. With Pro you see after each run which section cost you time.

No. Full text is the easiest route — type it or paste it from Pages, Word or an email — but you're not held to it while speaking: rephrase, and the app finds its way back. If you'd rather work without a written-out text at all, use the Freestyle cue card with bullet points and trigger words; that's part of Pro.

That's the normal case, not the exception. The app continuously compares what you say against your text — not word for word, but across the last few words. Rephrase, and it finds its way back at the next matching point. Nothing to do, nothing to confirm.

No. Recognition works from every word in your text, not only the marked ones. Accents are there for you, not for the app — they're more distinctive and therefore help you find your place again after a stumble, but the more you mark, the less your look-ahead tells you. Freestyle cue cards are the exception: there the app advances only via the trigger words you've stored.

Going forward, usually yes: as soon as you say a word that appears later in the text, the position moves there. For larger jumps, or if you double back, tap the word you're on.

A tap on any word resets the position, mid-speech, without opening a menu. It's the one gesture worth knowing before you stand in front of an audience with it.

No. Silence that wasn't planned as a timed pause counts only half against your target time — an interruption is rarely your speaking time. If it runs longer, pause the run and resume afterwards.

The text stays exactly where it is — it doesn't vanish and it doesn't jump. Keep speaking and tap yourself back into position. As a precaution, set the microphone orientation for lectern distance if the device isn't right in front of you.

No. During a run the app shows your speech full screen and keeps the display awake.

The interface comes in seven languages: German, English, Spanish, French, Italian, Dutch and Portuguese, with British English and European Portuguese as separate choices. You can deliver speeches in 27 languages across 54 regional variants. Language is set per speech — bilingual speakers keep their German and English speeches side by side.

This is where the regional variants earn their keep: an Austrian speaker picks an Austrian model, an Australian an Australian one — not merely "German" or "English". No recognition is perfect; a strong dialect, a fast pace, distance and room acoustics all have an effect. It matters less in practice than it sounds, because the app finds its way back via the last few words, and a tap resets the position anyway.

No. No voice profile, no training run, nothing to read aloud first — you write your speech and start. The only setup step is a one-time download if the model for a language isn't yet on your device.

Yes. Speech recognition runs entirely on your device: in airplane mode, in a parish hall with no signal, in a basement. The one prerequisite is that the language model has been downloaded once — after that you need no connection at all.

The model comes from iOS, not from us, and is downloaded once — after that it stays on your device permanently. On newer iOS versions the app handles it: in the speech's language settings, tap "Download model" and watch the progress. On older versions the route runs through the iOS settings; the app shows you the steps as soon as it notices a model is missing.

Stay on Wi-Fi while you do this — the download takes a few minutes:

1. Settings → General → Keyboard → Keyboards → Add New Keyboard
2. Add the language you need — the dictation language follows automatically
3. Back under "Keyboard": "Enable Dictation" must be switched on
4. In any app, Notes for instance, tap the microphone key on the keyboard and dictate a few words in that language

Step 4 is the one almost everyone misses: without dictating once, iOS won't download the model.

Not in any way you'd notice — we've measured it, and the difference is too small to matter while you're speaking. What differs is how you add a new language: that's more convenient on newer versions. Updating for recognition alone isn't worth it.

Chinese, Japanese, Thai, Cantonese, Arabic and Hebrew. Because the app is built on words separated by spaces and running left to right. Chinese, Japanese, Thai and Cantonese have no word boundaries in that sense; Arabic and Hebrew run right to left. This runs through the entire architecture rather than sitting in a setting — which is why we don't hold it out as "coming soon" either.

The design case is the device on the lectern, at reading distance. The further away it sits, the more recognition suffers — distance matters as much as accent, pace and room acoustics. There's a dedicated microphone orientation setting for exactly this position.

Through the microphone orientation: you tell the app the device is lying at a distance rather than being held in your hand.

The app uses wired microphones connected via Lightning or USB-C. It does not use Bluetooth headsets — in that case it keeps recording through the built-in microphone. And one common misconception up front: the microphone on the lectern is connected to the venue's PA system, not to your iPhone. The app always hears you through its own microphone.

Reverberation makes it harder — that's a genuine limit of on-device recognition, and we won't claim otherwise. What helps: the app doesn't need every word to be right; it continuously matches the last few words and picks up again at the next matching point. And if it does slip, a tap puts it back.

Ambient noise affects recognition, the same way distance and room acoustics do. Your speech doesn't depend on it, though: the text, the accents and the look-ahead stay on screen, the clock keeps running, and a tap corrects the position.

The presentation view is always dark, regardless of how your iPhone is otherwise set: light text on a dark background, built for exactly this situation. You can set the type size between 24 and 48 pt, with 32 as the default. The display stays awake throughout the run; brightness you control through iOS as usual.

Not with us. There is no SpeechMaestro server, no account, no sign-in — your speeches live on your device. If you want to sync between devices, that runs through your own iCloud account; the setting is off by default and is part of Pro.

Outside a run the microphone is not active. It's switched on when you start a speech and released when you leave the presentation view — while you're writing, sorting or reading a speech, the app isn't listening. During a run, iOS shows the orange microphone indicator throughout.

No. What you say is processed on the device to locate your place in the text, not recorded. There would be nowhere for it to go in any case: no server, no account, no analytics services. A take records the timing of your run, not a recording of it.

Recognition runs on the device, so the audio never reaches Apple's servers. Two things do involve Apple by their nature: your purchase through the App Store and — if you switch it on — syncing through your own iCloud account, where your speeches sit like any other iCloud data. With syncing off, no speech leaves your device.

No. We never get to see them — there is no server for them to go to, and no tracking or analytics services in the app. For board addresses, sermons and political speeches, that's often the actual reason people buy.

iPhone from iOS 17.6, portrait orientation; iPad from iPadOS 17.6, any orientation; Mac from macOS 14.6; Apple Watch from watchOS 11. Viewing and delivering on the Mac is free; editing on the Mac is part of Pro.

No — and that's a decision, not a gap. The app builds on Apple's on-device speech recognition, which is precisely what makes the privacy model and the offline capability possible.

You display the code on one device and scan it with the other; the receiving device needs a camera for that. Shorter speeches fit entirely inside the code — one scan, no account, no network. As a rule of thumb, a speech of around 300 words including a reference take fits inside the code, and considerably more without one. You don't have to work this out: the app picks the route itself and tells you which one it took when you send.

A QR code holds only so much data. If a speech doesn't fit, the code merely pairs the two devices and the text travels directly from one to the other — over Wi-Fi or Bluetooth, straight between the devices. They don't need to be on the same network, but they do need to be close together; they are anyway, if you're scanning a code. No server, no internet.

The app recognises from its origin that the speech is already on the receiving device and asks whether to replace it or keep both. Nothing is overwritten silently.

By QR code, in both directions. The Mac can both display and scan codes, using its built-in camera or an attached one. With Pro, iCloud sync through your own account is an additional route — more convenient, but not required.

Yes, and that's the default. Syncing is off unless you turn it on; if you never do, your speeches stay on the device they're on, and you move them between devices by QR code when you need to.

The title, time remaining, target time and a progress bar with milestones — so the phone can stay on the lectern. The bar's fill shows how far you are through the speech, its colour shows where you stand against your target. With Pro, a gentle tap also marks each milestone as you reach it, noticeable without a glance at your wrist.

Your speech text. The display is too small for it, and the whole point is not to look. You also can't control the run from the watch — no starting, no pausing, no ending. The phone leads, the watch follows. If the phone isn't in range, the watch stays on its waiting screen.

Markers along your way through the speech — preset at 25, 50, 75 and 100 percent of the text, individually switchable in the settings. The tap tells you how far you've come; whether you're on time is what the bar's colour tells you.

No, and it won't become one. Pro is a one-time purchase: €17.99 (US$19.99 in the United States). You pay once, or never.

You can create, edit and delete as many speeches as you like — and deliver every one of them in full. The delivery itself is never restricted: accent words, look-ahead, sections, timed pauses, a target time with a live pace display, all languages, QR sharing, import and export, and Apple Watch with progress and time remaining are all included. If you never want to measure, you never pay.

The analysis: the report after each run, section by section. Alongside that, saving takes and setting one as your reference, live pace measured against that reference take rather than a bare target figure, the Freestyle cue card, haptic milestones on Apple Watch, editing on the Mac, and iCloud sync. In short: the run is free, the result is Pro.

Yes. Pro is one purchase for all your Apple devices: buy it on iPhone and Pro is unlocked on iPad, Mac and Apple Watch too, as long as you're signed in with the same Apple Account.

Your purchase is tied to your Apple Account, not to the device, and there's nothing to pay again. In Settings, open a Pro feature; in the purchase window you'll find "Restore Purchases".