Long answer. This requires either WebCore pistons or a custom Smart App that takes your text and voice inputs, gets the sound, and has the sound sent to the speaker for execution. I do not know if one exists, but it sounds like fun.
WebCore users: Does the WebCore textToSpeech function allow entry of a voice (second parameter in command line after the text). If so, I believe this is a no brainer. I have tested the command to my Samsung R1’s, and the device handler accepts the voice parameter (and actually works).