What’s new
New features and pricing changes, written down here.
- 1.1.1New
The Windows app is here
The desktop app now downloads on Windows 10 and later. There is no signing certificate yet, so the first install shows one warning — we have written down how to get past it.
The desktop app, until now macOS only, is available for Windows. It needs 64-bit Windows 10 or later, and the installer is 78 MB. Same account, same notes, same pricing — the app only handles recognition and display, while notes, speaker labels, AI minutes and the audience QR all read the same data as the web.
One thing is different: signing. A Windows code-signing certificate is not in place yet. So the first time you run the installer, Windows shows “Windows protected your PC” and hides the Run button behind “More info”. That is the absence of a signature, not a problem with the file. The download page spells out the two clicks.
All five global shortcuts work on Windows: toggle the captions, nudge them a line up or down, and make them bigger or smaller.
- Windows 10 or later · 64-bit · 78 MB installer
- First install: SmartScreen warning → “More info” → “Run anyway”
- All five global shortcuts work (Control+Shift+S / ↑ / ↓ / + / -)
- 1.1.0
Desktop app — the recognition window is gone in AI Presentation mode
One app handles everything, with no separate Chrome window. And you can drag the caption window wherever you want it.
Presenting from the desktop app used to open a separate Chrome window for recognition — a limitation of the browser engine we were using. AI Presentation mode does not use that engine, so it does not need the window. Now there is one app, captions cannot stop because a window was closed, and a missing Chrome cannot stop you from starting.
Besides picking a position from the grid, you can now drag the caption window where you want it and resize it from the corner — useful for keeping it off a logo or a lower third. Set it up on your laptop and it lands in the same place on the projector.
The caption window can show the original and the translation together, stacked one language per row or side by side.
- AI Presentation: no recognition window, no Chrome, no sign-in step
- Subtitle-only Presentation mode still uses Chrome — that is why live subtitles are free
- If Chrome does cause trouble, one tap switches the note to AI Presentation
- Place the caption window by hand; tap the grid to return to a preset
- The global shortcuts (Control+Shift+S and friends) are unchanged
- 1.1.1
Presentation and AI Presentation are now two separate modes
No more checkbox asking whether to enable AI voice. Pick the mode up front and the rest follows.
Previously there was an "enable AI voice" checkbox inside presentation mode, and once you ticked it you had to choose again which languages got audio. Two decisions for one intent, both buried near the bottom of a setup screen.
Now you choose between "AI Presentation" and "Presentation" when you create the note. In AI Presentation the translation languages you pick *are* the spoken languages — your audience hears them on earphones. Presentation sends subtitles only.
Both modes sit at the top of the list, and AI Presentation is the default. The button that silences the voices mid-talk and drops back to subtitles is still there.
- AI Presentation runs up to 4 languages at once, Presentation up to 8 (each voice uses one of the presenter’s network connections)
- The supported languages are the same in both modes — all 34 work with AI voice
- Tell us each speaker’s gender and tone and the voice is matched to them
- On the audience screen, a 🎧 badge marks the languages you can listen to
- 1.1.1
Fixes to the audience language picker
Languages that would never be translated no longer appear, and the lower entries are reachable when there are many.
A talk running Korean→English could show Japanese and Chinese on the audience screen, and picking one left you on "Translating…" forever. The languages the presentation screen was actually producing and the languages offered to the audience were being read from two different places. They are now written down once when the talk starts, and the audience list shows only those.
With many languages the list also overflowed the screen with no way to scroll to the bottom entries, and scrolling the page left the buttons on a white background. The heading is now pinned and only the list moves.
- Only languages the talk is actually translating into are listed
- A 🎧 badge on languages you can listen to; a "floor language" note on the original
- The list scrolls properly and keeps its background, however many languages there are
- An older line whose translation never arrived shows the original, dimmed, instead of spinning forever
- 1.1.0
AI voice interpretation — your audience listens on earphones instead of reading
An AI speaks what the presenter said, in whichever language each attendee chose. If the presenter switches language, the listener keeps hearing theirs.
Until now, anyone who joined through the audience QR code spent the hour reading subtitles on a phone. An hour spent looking down at your palm is not an hour spent at the talk. Press “listen” and you can hear the interpretation on your own earphones while watching the stage.
If the presenter opens in English, switches to Japanese, and a second presenter continues in Korean, an attendee who chose Korean keeps hearing Korean throughout. With several presenters, each is assigned a distinct voice that stays the same all session, so you can tell who is speaking by ear alone.
No real person’s voice is cloned. Each speaker is assigned one of the provider’s stock voices, chosen so they are easy to tell apart.
- Switched on per language from the presentation screen — up to four at once
- Leaving a language on costs nothing until somebody actually listens to it
- One listener or a thousand, the deduction is the same
- Attendees who pick a language without AI voice get their phone’s own voice reading the subtitles, at no extra cost
- Still no account and no app install for the audience
- 1.1.0
We fixed how pricing is described — “one credit, one second” is gone
What gets deducted depends on what you asked for, so stating “1 minute = 60 credits” as an identity was not accurate.
The pricing page long stated “60 credits = 1 minute” as though it were an identity. Strictly, it never was — a talk with nine subtitle languages has always spent 156 credits a minute. The gap was just small enough to explain in a footnote.
AI voice interpretation made that gap too large for a footnote. So the page now says plainly what a credit always was: a unit of work, not a unit of time.
This is not a price rise. Nothing about the existing subtitle features deducts any differently. Only the new AI voice costs what the new AI voice costs.
- Subtitle languages are unchanged: +20% each
- A language with AI voice on uses two extra minutes per minute (+200%)
- The pricing table now shows deductions for voice-enabled setups too
- The free 40 minutes, the pass prices and their validity are unchanged
기능 문의나 요청은 develop@amuz.co.kr 로 보내 주세요.