dbiglari

That is terrific output. Your output sounds similar to the audio samples https://google.github.io/tacotron/ without the post processing network and the vanilla seq2seq. The buzz is present, but a voice is...