My Speech Server is all the way thug
So far, I'm throwing a few ideas around to really make my upcoming code camp presentations good. I just ask myself “What would Casey Chesnut ” do and then I get good ideas. So I was playing around with it, and started putting in Lyrics to Fifty Cent Rhymes as my prompts. You DEFINITELY need to use Fifty's voice and load them a la Prompts db for this. If you dare try to put the lyrics in and roll via TTS, it's ugly. TTS Can Not rap at all. I mean zilch. I could rap better and that's saying a lot. Heck, I think John Ashcroft could rap better than TTS. But the funny part is that it does pronounce the words correctly. Ostensibly the best line of the bunch was:
<<I'm a boss, but nigga's never show no respect
I catch em slippin, I have em tongue kissin my tech- >>
The whole effect is just totally lost when it's dispensed via TTS.
Anyway, I parsed enough files from the massacre to get a killer prompts db. Building the grammars to support it though - it's UGLY. I could probably spend all month because of all of the different ways words are pronounced.
If you want a good chuckle though, go find some Geto Boys or Body Count and pop the lyrics in a QA control. I can just about guarantee that you'll laugh your a55 off.