As I said, I am aware of the fact that these two are different things and the fact that the level of complexity is different - no need to convince me of something I already agreed on
As to the rest of what you wrote, I'll risk being off-topic for one more reply, just to make clarify some things on the speech recognition thing - that will be my last off-topic reply on the matter here, so don't worry. The whole speech recognition stuff is much older than Cortana in MS (I even have some free source code on the matter for a different program, that's using Microsoft Speech Recognizer 5.x from the 2000 era - check , if you are curious about it). Bottom line, the stuff is exposed through an API since ages ago, and nothing more complex needs to be done in a program that's using it other than check out for the API's "translation" of the spoken content.
The fact that you have few programs using this capability is not because it's extremely complex for that program to do (because nearly everything is done by the MS API), but because it's highly unreliable, for the reasons you alredy mentioned (colds, hangovers, different grammars, languages, etc). So basically what I'm saying is that while I understand you're not interested in this (and for good reason, aka the poor reliability), it isn't that hard to accomplish - it's just about using a specific API.
Coming back to being on topic (sorry for the clarifying detour), I look forward to the moment when your plugin will be "official" (i.e. included in the Rainmeter distribution). It may take longer or it might never happen, but it is as valid of a candidate as any other plugin out there.