Directly Modeling Voiced and Unvoiced Components in Speech Waveforms by Neural Networks

   Abstract