View Single Post
Benson's Avatar
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#10
Originally Posted by geneven View Post
The Windows-based synthesized voices I use for text to speech have no big problem with street names, though they might get some wrong, as might a human. Listen to the voice on www.iht.com, for example, or the samples on www.nextup.com. The good voices sound good. Linux just doesn't have good voices. A decent tts program such as textaloud also lets you modify pronunciations if you want.

I think you're misunderstanding here. The contrast is between recorded phrases (can't say street names, sounds good) and speech synthesis (gets street names, doesn't sound as good).
While it's true that the voices for speech synthesis vary hugely, even the best aren't as good as recorded phrases.