There's very little demand for that, and it requires a lot of horsepower. It also tends to be imprecise and error-prone.
Most operating systems, including Windows, allow for third-party drivers to be installed to support just about any device. If someone writes drivers that allow a microphone and speaker to be substituted for a keyboard and screen, Windows will support it just fine. Getting voice input to work is hard enough, but converting all the visual information in Windows to audible output is a Herculean task, and requires skills and techniques that nobody actually has right now.
Disabled people don't make anyone lots of money. Even so, many companies, including Microsoft, spend more money accommodating them than such customers bring in.
It may, or it may not. I've never seen any proof that speech input and output is in any way superior to the current arrangement. They are handy when one cannot type or see, but if one can type and see, they aren't that useful, except as novelties.
Microsoft doesn't build specialized hardware or drivers for such hardware, nor is it in the habit of stealing such things.
They are lazy about producing software for the disabled because they don't see any money in it, and they are not operating as charities. The market for such specialized hardware and software is too small to allow the costs of development to be recovered in sales.
-- Transpose mxsmanic and gmail to reach me by e-mail.