1. 23 10月, 2019 1 次提交
    • 1024的传说's avatar
      change rnn_search to seq2seq; change vae_text to variational seq2seq (#3734) · b034390c
      1024的传说 提交于
      * re-commit seq2eq example: rnn search
      
      * add new_seq2seq examples: rnn_search and vae_text
      
      * modify dataset description in vae_text/README.md
      
      * modify vae_text/README.md and vae_text/download.py
      
      * update vae_text/infer.sh and vae_text/README.md
      
      * add multi-gpu support for rnn_search; fix windows support for rnn_search and vae_text
      
      * remove parallel argument in args.py
      
      * remove hard codes
      
      * remove hard codes
      
      * update download.py for windows download
      
      * fix model path for windows in vae_text/infer.py
      
      * fix python2 utf-encoding
      
      * change rnn_search to seq2seq; change vae_text to variational seq2seq
      
      * change rnn_search to seq2seq; change vae_text to variational seq2seq
      
      * mv variational seq2seq to variational_seq2seq
      b034390c
  2. 28 6月, 2019 1 次提交
  3. 28 4月, 2019 1 次提交
  4. 04 4月, 2019 1 次提交
  5. 02 2月, 2019 1 次提交
  6. 01 2月, 2019 1 次提交
  7. 01 2月, 2018 1 次提交
  8. 16 1月, 2018 1 次提交
  9. 14 6月, 2017 1 次提交
  10. 12 6月, 2017 1 次提交
    • X
      Refactor whole data preprocessor for DS2 (re-design classes, re-organize dir,... · 70eb4001
      Xinghai Sun 提交于
      Refactor whole data preprocessor for DS2 (re-design classes, re-organize dir, add augmentaion interfaces etc.).
      
      1. Refactor data preprocessor with new added class AudioSegment, SpeechSegment, TextFeaturizer, AudioFeaturizer, SpeechFeaturizer.
      2. Add data augmentation interfaces and class AugmentorBase, AugmentationPipeline, VolumnPerturbAugmentor etc..
      3. Seperate normalizer's mean and std computing from training, by adding FeatureNormalizer and a seperate tool compute_mean_std.py.
      4. Re-organize directory.
      70eb4001