deepsound-project/samplernn-pytorch

Issues

• YouTube-Mix dataset possibly leaks training data to validation and testing

  #47 opened a year ago by markschoene
  0

• CuDNN while generating is now available (in master)

  #6 opened 7 years ago by qbx2
  4

• "Could not find a version that satisfies the requirement torch==0.2.0.post3"

  #26 opened 6 years ago by Bryncodeswithfriends
  4

• Working training in colab but no sound

  #40 opened 4 years ago by pabloriera
  7

• Which torch version to get?

  #45 opened 3 years ago by Flesco
  0

• How do you sample from the model you created?

  #44 opened 4 years ago by Randy-H0
  0

• How do you load models?

  #43 opened 4 years ago by Randy-H0
  0

• OSError: [WinError 123] The filename, directory name, or volume label syntax is incorrect:

  #42 opened 4 years ago by Bjerken
  0

• Why is the input data not normalized？

  #41 opened 4 years ago by MaiSpace
  0

• Explanation of `frame_sizes` and `ns_frame_samples`

  #17 opened 7 years ago by simopal6
  3

• Why is overlap length in dataloader needed?

  #25 opened 6 years ago by williamFalcon
  1

• training stops in colab

  #34 opened 6 years ago by ak9250
  4

• Successful implementations?

  #31 opened 6 years ago by Stafla
  1

• Conflict between PyTorch version and CUDA version

  #38 opened 5 years ago by Sinnerboy89
  1

• Thank you for making this. Is there an easy way to get this to run on TPUs in Colab?

  #39 opened 5 years ago by Tylersuard
  0

• Bad amplitude normalization

  #21 opened 7 years ago by Cortexelus
  3

• Getting runtime error for sizes of tensors not matching

  #36 opened 5 years ago by masheendream
  1

• No module naimed trainer

  #35 opened 5 years ago
  2

• sample generated is noise

  #29 opened 6 years ago by guozixunnicolas
  1

• MIDI file as input data

  #30 opened 6 years ago by guozixunnicolas
  0

• Loss increase dramatically after 60000 iterations

  #33 opened 6 years ago by qiuqiangkong
  0

• bug in hidden state?

  #32 opened 6 years ago by qiuqiangkong
  1

• How to use model for generation

  #28 opened 6 years ago by manoharvhr
  1

• download dataset from youtube

  #27 opened 6 years ago by guozixunnicolas
  1

• Talkin' 'bout my generation

  #8 opened 7 years ago by lonce
  5

• Division by Zero when training

  #12 opened 6 years ago by LukeB42
  10

• What does n_frame_samples represent?

  #23 opened 6 years ago by williamFalcon
  13

• Convolution dim shuffling

  #24 opened 6 years ago by williamFalcon
  4

• Reconstructing specific wav file

  #22 opened 7 years ago by JSai23
  1

• README needs to have PyTorch version updated

  #20 opened 7 years ago by convoliution
  0

• SampleRNN as audio feature extractor

  #18 opened 7 years ago by iariav
  1

• Sizes of tensors mismatch in dimension 0

  #19 opened 7 years ago by niuqun
  1

• No matching distribution found for torch==0.2.0.post3

  #15 opened 7 years ago by HandsomeDevilv112
  2

• CUDNN_STATUS_NOT_INITIALIZED

  #16 opened 7 years ago by HandsomeDevilv112
  1

• Generation is always the same

  #14 opened 7 years ago by davidmrdavid
  1

• pause/resume

  #13 opened 7 years ago by freekatet
  2

• Training time?

  #11 opened 7 years ago by turian
  0

• ZeroDivisionError: division by zero

  #10 opened 7 years ago by turian
  1

• ffmpeg is not present in Ubuntu

  #9 opened 7 years ago by turian
  1

• Why is prev_samples = 2 * dequantized ?

  #7 opened 7 years ago by qbx2
  3

• https://github.com/soroushmehr/sampleRNN_ICLR2017/blob/master/models/three_tier/three_tier.py#L496

  #2 opened 7 years ago by skaae
  7

• quantizer

  #1 opened 7 years ago by skaae
  2