Hybrid phoneme based clustering approach for audio driven facial animation.