Skip to content

Desktop App v0.3.2 (superseded by v0.3.3)

Choose a tag to compare

View all tags
@Iyanuoluwa007 Iyanuoluwa007 released this 12 Aug 12:47
· 11 commits to main since this release
desktop-v0.3.2
32575d6

Signs what your whole computer is saying, not just your browser tabs. It reads Windows 11 Live Captions, which transcribes system audio, so it works with video calls, media players and any other desktop application.

Update from v0.3.0 or v0.3.1. Those builds had a bug that made the controls unclickable.

All downloads

Product Platform Version File
Desktop App Windows 11 v0.3.2 Signlytic AI Setup 0.3.2.exe (99 MB)
Desktop App macOS upcoming not yet available
Chrome Extension Chrome v0.3.6 signlytic-extension.zip (23 MB)

Fixed in v0.3.2

The buttons work again. In 3D mode the avatar was drawn on a canvas that
sized itself in fixed pixels and spilled out of its area, ending up on top of
"Start Live Captions" and the text box. That is why the avatar appeared down
beside those controls and why neither could be clicked: the canvas was over
them, taking the clicks. The canvas is now pinned inside its own area and can
no longer cover anything.

From v0.3.1

"Start Live Captions" opens Live Captions for you if it is not already
running, then begins signing. An already-open Live Captions is left alone.

From v0.3.0

It overlays your desktop rather than sitting in a window. Transparent and
frameless, always on top, drawn as a translucent panel.

Five placements. Snap it into any corner or leave it floating; your choice
is remembered between runs.

The caption reader survives failures. It no longer stops on a transient
Windows UI Automation hiccup, restarts itself if it does stop, and reports a
readable reason. This is the fix for "sidecar exited (code 4294770688)".

Installing

Run the installer. It is unsigned, so Windows SmartScreen will warn on first
run: choose More info, then Run anyway.

Requirements

  • Windows 11 22H2 or later, for Live Captions. Its first launch asks you to
    accept terms and download a speech model, which is a one-off Windows step.
  • An internet connection: gloss translation and sign data are fetched from
    the Signlytic API.

Known limitations

  • In 3D mode the avatar's hands sit lower than they should. 2D is the default
    and is unaffected.
  • No system tray or auto-start yet.

macOS

The Mac app is upcoming. The app already runs on macOS through manual text
entry; what is missing is caption capture, because macOS exposes captions
through the Accessibility API rather than the interface used on Windows.

Third-party assets

The 3D avatar models are Mixamo characters and are not licensed for
redistribution by this project. They are used for rendering only and are not
offered as standalone downloads. Obtain your own from https://www.mixamo.com
under an Adobe account.

Assets 3
Loading