It's truly amazing what having to think before you speak can accomplish.
If I remember correctly he was stuck on some long discontinued TI speech synthesis chip. I remember the initial story when they first did it in particular because I also was playing with a similar part at the time in High School.
Finding parts and people has to be getting difficult.
At the same time I can completely understand him not wanting to upgrade his system, from the GUI he's used very successful for 25 years or Voice that has now become that trademark of Steven Hawking's.
Well I am thinking we should be able to emulate that whole system including the speech syntheses. Worst case it can be done as a series of recordings from the original chip.
Maybe do it as a Kickstarter project or something, do the whole thing open source. Surely there must be others with this problem.
I can imaging upgrading him to AR goggles with eye tracking, that's all off the shelf today. Then in to a little Mini PC or ARM board, and nothing but software.
With something like that we should be good till we get direct brain interfaces and quantum computing wrist watches in another 25 years, if he still even needs it.