Supplemental data for “C-RNN-GAN: Generative adversarial training of sequence models with continuous data”.
Music generated by the model.
- sample11.[mid]
- Original filenameout-20161208-deep_large_d_generate_meta_large_random_inputs_3tpc-18-2016-12-09-17-06-07.mid
- Feature matching, freezing, 3 tones per RNN cell, larger D (500 hidden units, LSTM depth 3), generate metadata
- 18 epochs training
- sample10.[mid,mp3]
- Original filename: out-20161114-3tonespercell-freezing-82-2016-11-15-12-49-34
- Feature matching, freezing, 3 tones per RNN cell
- 82 epochs training
- sample09.[mid,mp3]
- Original filename: out-20161107-rnn-gan-lr.1-featurematching-65-2016-11-07-16-43-04.mid
- Feature matching
- 65 epochs training
- sample08.[mid,mp3]
- Original filename: out-20161107-rnn-gan-lr.1-featurematching-52-2016-11-07-16-07-19.mid
- Feature matching
- 52 epochs training
- sample07.[mid,mp3]
- Original filename: out-20161107-rnn-gan-lr.1-featurematching-50-2016-11-07-16-01-40.mid
- Feature matching
- 50 epochs training
- sample06.[mid,mp3]
- Original filename: out-20161107-rnn-gan-lr.1-featurematching-40-2016-11-07-15-33-44.mid
- Feature matching
- 40 epochs training
- sample05.[mid,mp3]
- Original filename: out-20161107-rnn-gan-lr.1-featurematching-39-2016-11-07-15-30-11.mid
- Feature matching
- 39 epochs training
- sample04.[mid,mp3]
- Original filename: out-20161107-rnn-gan-lr.1-featurematching-29-2016-11-07-14-41-20.mid
- Feature matching
- 29 epochs training
- sample03.[mid,mp3]
- out-20161105-rnn-gan-lr.05-featurematching-108-2016-11-07-00-28-25
- Feature matching
- 108 epochs training
- sample02.[mid,mp3]
- Original filename: out-20161105-rnn-gan-lr.05-featurematching-105-2016-11-07-00-20-17.mid
- Feature matching
- 105 epochs training
- sample01.[mid,mp3]
- Original filename: out-20161105-rnn-gan-lr.05-featurematching-103-2016-11-07-00-15-23
- Feature matching
- 103 epochs training