I think its done by overlaying pitch-changed copy of the voice on top of the original voice