Getting the right text-to-speech voice is probably more important then the software that runs it. In a study looking at the quality of TTS voices found that there were three quality levels, high, moderate, and low (Cunningham, 2011). A high quality TTS voices is simulate to that of a humans. Many of the companies that create TTS voices provided free demonstrations of the voices.
Past research has also shown that it takes time for a person to "acclimatize" to a new text-to-speech voice. Reynolds, Isaacs-Duvall, and Haddox, (2002) found that it took a listener about 30 minutes listing to a low quality TTS voice before there auditory processing had attuned to the new voice.
It is very important that users listen to the a varitey of TTS voices and chose one that is most human sounding and that they understand. With over 80 languages have TTS voices and many different access, there is quit a selection of voices at this time.
For a list of voices please see Nextup for voices that can be download onto your windows based computer.
Special Consideration: Workflow
Exact prices change frequently, which is why only approximate ranges are listed.