I've noticed something weird about Japanese TTS systems. Instead of just having generic voices, each voice has a character, similar to the way characters work in singing synthesizers, where a community ends up forming around each voice.
I'm not entirely sure I understand why.