Estadístiques de Towards automatic extraction of prosodic patterns for speech synthesis