Abstract: We provide a speech coding scheme employing a generative model based on SampleRNN that, while operating at significantly lower bitrates, matches or surpasses the perceptual quality of ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results