device independent layer for speech synthesis