Directly Modeling Speech Waveforms by Neural Networks for Statistical Parametric Speech Synthesis

   Abstract