A New Approach to Voice Authenticity