It is already possible to cut / paste and even morph a person's voice to some fair degree! I am a casual musician (play on weekends) and the software / hardware for manipulating and morphing audio is becoming quite powerful.
Without getting too techno, there are things called
formants which gives the word its individual characteristic timbre in terms of male-female, old-young etc.
For example I have a hardware based vocal harmonizer called a
Voiceworks which not only allows multiple versions of my voice to be pitched higher or lower, but can take on characteristics not inherent to my actual timbre.
How much more sophisticated would such manipulations be in the hands of the National Security Agency or the Canadian Security Intelligence Service (well maybe not us Canadians) or some covert black ops agency? Enough to fool someone on the telephone?