Does anyone have experience determining performance levels for wav generation.
I am very interested to know on a windows machine if there is substantial performance gains by switching to either the handheld voices or even the telephony voices.
*Does quality of voice effect performance?*
Secondarily I would like to know if the number of ports on a multicore or multi-proc machine scales linearly. Few applications do - which means there is an ideal number of ports for each machine. If Cepstral has done any testing on this - I would love to hear their reccommendations. My limited experience with this voice engine show it pegging a single core VM to the roof given only a single port. I would also like to know whether adding another port would just a bad situation worse?
*Are there reccommendations on the number of ports for a particular CPU setup?*