I read Windup Girl. Ship breaker I did audio and you're right, he sucks. There are worse, though. I strongly disagree about text-to-speech. A decent narrator is a thing of beauty. Roy Dotrice's reading of Game of Thrones is sublime, and the Hunger Games are read by Carolyn McCormick, who I'm sure has smaller hair these days: Text to speech will never get there. The last financial book I listened to had a narrator who didn't know how to pronounce "arbitrage." If humans can't even pull it off, robots will always be infuriating. And I own a professional text-to-speech engine. The handwork necessary to make things passable is pure drudgery.
It'd be nice to have an awesome narrator for any book, not just the lucky ones. One can dream though... And hey, there are some phenomenal vocaloids out there in the world of music: I have no depth to my knowledge of how much twiddling it takes to produce even those 8 minutes, though.
I'll bet Vocaloid is less of a pain in the ass when you know Japanese. Yamaha had the Vocaloid team at NAMM. Not last year, but the year before. They were... otaku. They also showed how you could construct a song in like no time at all... in that same way that Avid demonstrates you can make a rap track in Pro Tools that if you haven't done it 8 times a day for 5 months would actually take you two afternoons. As it is even the Japanese, who have pushed a couple Vocaloid tracks to the top of the charts, still use voice actors for game and movie talent. Vocaloid is a thing but it's not going to replace human speech any time soon.