Fig. 2From: Developing a unit selection voice given audio without corresponding textSteps followed for acoustic modelingBack to article page