r9y9/tacotron_pytorch

Issues

• about validation

  #23 opened 5 years ago by ArtemisZGL
  0

• why turning off dropout of decoder's prenet make a serious performance regression

  #20 opened 6 years ago by zwlanpishu
  2

• Trained Model

  #22 opened 5 years ago by madaanpulkit
  0

• Masked loss function

  #21 opened 6 years ago by njellinas
  2

• retraining from the checkpoint on GPU fails

  #18 opened 6 years ago by saikrishnarallabandi
  4

• BahdanauMonoAttention cannot work well

  #8 opened 6 years ago by zhbbupt
  2

• About BatchNorm

  #19 opened 6 years ago by mazzzystar
  3

• Where is your checkpoint_step720000.pth?

  #17 opened 6 years ago by pzSuen
  2

• ModuleNotFoundError: No module named 'text'

  #12 opened 6 years ago by dza6549
  5

• why is_end_of_frames can detection the end frame in test phase?

  #15 opened 6 years ago by wangjunchao1118
  4

• GPU inference time?

  #13 opened 6 years ago by PetrochukM
  2

• Implementation of Bahdanau attention is possibly different from paper?

  #10 opened 6 years ago by rafaelvalle
  9

• Update model

  #2 opened 6 years ago by r9y9
  37

• Try max_norm=1 for Embedding layer

  #4 opened 6 years ago by r9y9
  1

• Alignment mask

  #16 opened 6 years ago by pravn
  1

• transpose before conv1d_banks？

  #14 opened 7 years ago by ZuoChenFttS
  1

• Why sort within batch

  #9 opened 7 years ago by tivaro
  1

• Checkpoint file and dropout on eval.

  #11 opened 7 years ago by pbaljeka
  2

• Muli-speaker embedding

  #5 opened 7 years ago by r9y9
  1

• How are linear targets being passed to the model?

  #7 opened 7 years ago by 7404N
  2

• Double denormalization?

  #6 opened 7 years ago by qbx2
  2