Emacspeak at Twenty: Looking Back, Looking Forward
emacspeak.blogspot.com
emacspeak.blogspot.com
When Raman started working on Emacspeak in 1994, most blind computer users, including him, were using screen readers for MS-DOS. Naturally, these only worked in text mode. They intercepted BIOS interrupts and/or polled screen memory to find out what was on the screen. So naturally, these screen readers didn't have access to any information about the structure of what was on the screen, just the text and colors. They used various heuristics to determine what to speak in a given situation, but this was far from optimal IMO. I dare say the same applies to Linux text-mode screen readers, specifically Speakup.
Today, there are multiple excellent screen readers for Windows, both commercial and free. The most popular commercial screen reader is JAWS for Windows; the best and most popular free one is NVDA. Apple provides a good screen reader called VoiceOver on both Mac and iOS. On Android there's TalkBack, and for GNOME, there's Orca.
The primary way of providing access to GUIs is through accessibility APIs such as Microsoft Active Accessibility and UI Automation on Windows. These APIs expose GUI widgets and containers as a tree of objects, with attributes such as name, role (e.g. button, list view, editable text), state (a set of flags such as focused, selected, checked, etc.), value, location, and more. So there's much more information about the structure and meaning of what's on the screen; this is practically a necessity for GUI screen readers, as opposed to text-mode ones.
Some GUI screen readers, such as JAWS, NVDA, and Orca, also provide ways to associate scripts or plug-in modules with specific applications, to add application-specific behavior. The ability to modify screen reader configuration settings on a per-application basis is also common (as it was in DOS screen readers).
I don't have an informed opinion of my own about Emacspeak versus the current crop of GUI screen readers, since I don't use either full-time. But based on my observation of the blind people I happen to know, including blind programmers, Emacspeak doesn't seem to be very popular even among blind programmers; the ones I know mostly use Windows and NVDA.
Anyway, I hope this comment will help to provide some context for the OP. It's not often that I see a post about assistive technology hit the HN front page.
Here's a description of the process of sending an email by the creator of Orca, the GNOME screen reader.
Also I wonder if vim or IDE's like IntelliJ provide comparable accessibility. I'd hope they do.
It doesn't mess with your scrolling or use unreadable color combinations. It doesn't use crazy fonts or table layouts.
What's unreadable?
Curious question: As a blind programmer - has the 'JavaScript everything' approach to the web made your personal content consumption easier or harder?
I briefly crossed paths with some blind users back in the late 1980's back when there was no heavily established GUI for the sighted masses (i.e. Windows 3.11). I'd be interested to know, compare and contrast how easily 'consumable' present cyberspace is versus the command-line era.
I'll take dog tiles vs pop-up ads or ads embedded in the text.
Modern web sites often feel like flash hell from the 90s ( http://www.penny-arcade.com/comic/1999/03/03 )