I have never been fond of screenreaders. For a very intuitive reason - they try to ADAPT the user experience of the sighted for us. And that, you see, is not how user experience for the differently abled should be designed. Every user interface should be designed with that user in mind - First. The idea of adapting the user interface of one category of users to another category of users is prima facie counter intuitive.
With the mobile phones and touch phones, that problem becomes worse. A mobile phone, as we know, can become a real friend with an audio guide like Siri. Instead, my accessibility talkback feature first trains me on how to touch the various areas of the screen to reach the app i want.
From a UX perspective, thats all wrong. I don't need to know where the apps are on a screen. I just need to open the one i want. Which means i should be able to tell the device what i want, and the device should be able to meet that need.
We have created touch screen laptops for our sighted users. If that's possible, tell me again why complete voice navigation based devices appear so unimaginable.
Here, then, is UX Design 101, as applied to differently abled users that we work with:
1. Voice Based Recognition and authentication - this includes special training on local accent customisation.
2. Voice based program triggers - You do realise that actually, we only need a screen to show someone sighted our work? So the screen should not be the primary trigger for program activation, change and closure. The voice command or tactile buttons should be.
3. Intra Program actions - this can be website usage, using office or personal productivity software, or playing games. The intra program actions can easily be designed so that they are voice controlled. We should be able to review our work using playback.
I highly recommend the native built in talk back feature of MS Excel (not sure if its still there in the new MS Office)
4. Braille? Maybe: We may or may not know Braille. That's all.