It uses an entirely different synthesis method than UTAU's approach, and the voice database looks to be an encrypted/compiled file, much like Vocaloid.
