Automatic Speaker Recognition using LVQ with 3 Prosodic Features