An well-engineered solution for this problem domain would be a clean factorization of all the stages with well-designed text-based IR formats between each stage, all data driven rather than massive monolithic codebase.
-
-
Show this thread
-
Especially important in cases where you're interested in speech synthesis that isn't TTS.
Show this thread
End of conversation
New conversation -
-
-
Surely there is a good scheme toolchain now that does support modules? Festival is pretty old though.
-
Probably. Bad module system architectures come in all languages. I was more reacting to their philosophy bs about good sw engineering juxtaposed with that.
-
Well, I'd charitably call it aspirational in a world of hard choices. But, have you seen CMU Sphinx? It's all relative.
-
Seems like it's an institutional problem and "we're doing slightly better on this project than that one" doesn't mean a lot..
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.