Not only is it good enough for mainstream adoption, it is mainstream. It's available in cars and practically every GPS device these days. Any time you call your credit card company, you deal with TTS (plus a fair bit of speech-recognition).
As for robots, sure we don't have robots of the kind in the images shown, but the machinery we use to build skyscrapers in a fraction of the time it took 100 years ago is quite amazing. The heavy-earth machinery, cranes, railway-track building machines etc don't look like robots, but they do the same stuff.
I'd be interested to see what percentage of random dialed phone numbers end up connected to text-to-speech "if you want sales, press 1 now. If you want support, press 2 now, for all other queries, press pound or just wait." type of responses.
(Hmmm, a free $30 credit Twilio account and their "detect answering machine" API call could generate a sample of 3000 random phone numbers pretty quickly...)
I'm not sure what your definition of mainstream is, but I commonly see text-to-speech listed as an advertised feature of mainstream consumer products.
TTS's dirty little is that it's actually much faster to read something yourself than have something read to you, assuming you are fluent in the language. It's not a failing of TTS, it's just people not properly considering the use of technology. The same goes for the roomba and the programmable microwave. It makes much more sense to have small single-purpose robots that function well, than a single multipurpose humanoid robot.
Roomba, on the otherhand, is a pathetic, if you compare it to visions.