Investigating the Use of Mixed-Units Based Modeling for Improving Uyghur Speech Recognition