Open sandipjedhe opened 5 years ago
I am thinking of it, but due to my current workload this has to wait :-(
@sandipjedhe I already implemented prosody embedding without GST with nvidia/tacotron2. You can modeling f(ref audio, speaker id, text) -> target audio using https://github.com/Yeongtae/tacotron2/tree/prosody_speaker_embedding_test.
@ErnstTmp and @Rayhane-mamah Is anyone combined tacotron2 and gst.
Or anyone pls comment here