ViSQOL: an objective speech quality model

  Abstract