[Proof of Concept]: Vosh - a third-party screen-reader for the Macintosh

2 min read Original article ↗

Forum

App Development and Programming

While I'm no longer interested in actively working on this, the Element library target in the current code is quite robust and may be of use to others at least as an example of how to safely interact with the accessibility infrastructure on macOS. Some features made available by the accessibility infrastructure are not wrapped by my code yet, and one of the things that I definitely want to do is reverse-engineer the custom rotors since there's no public API to consume those. If and once I manage to reverse-engineer the custom rotors, I will update this post again with a video of the latest code in action.

The source code of this project can be found on my GitHub repository, and at the time of this edit, is fully conformant with macOS 26 and Swift 6.2 with its complete structured concurrency model.

If I ever manage to find enough time, one of the things that I'd like to do just for fun at some point in the future, is to try training a custom machine learning model to deal with all accessibility quirks and even provide proper content recognition in totally inaccessible applications, though this is not a promise since I have a lot on my plate right now.


Below is the main paragraph from my original post to this thread before this edit today, 2026-01-12:

After getting fed up with the general neglect of MacOS accessibility from Apple, and having wanted to work on something meaningful for quite some time, I decided to attempt something that for some reason nobody seems to have tried to do before: write a completely new screen-reader for that platform. This isn't an easy task, not only due to the amount of work required to even get close to matching a mature screen-reader in terms of functionality, but also because Apple's documentation for more obscure system services is nigh on non-existent. Despite that, and since I've already overcome a lot of hurdles that I thought to be show stoppers, after a single week of work I already have something to show in a very embryonic stage of development. The idea is to gage the interest of the community in a project similar to NVDA for the Mac to be worked on in the coming years and to which other people can contribute.