client had JAWS, and the JAWS installer read itself to me while it was installing - I was rather impressed. Had to reboot, and it never came up after that - assuming it was related, but never figured it out.
The NVDA screen reader did work, as well as the windows voice assistant thing itself. both would generate audio in the workspace, and I was hearing it over my el capitan local setup.
Don't have time now, but hopefully it will be resolved at some point. NVDA was 'good enough' for the testing I was doing, and it was pretty slick to know the audio was streaming and working without any extra configuration. Probably old hat for some folks, but I'd never experienced that before. :)