This repo is use tf.estimator for wavenet vocoder. Input audio is raw or mu-law quantize.
- tensorflow v1
- numpy
- librosa
- scipy
- sklearn
- tqdm
https://kokeshing.com/wavenet/
- gaussian wavenet(Implemented)
- global conditioning(Implemented)
- 2dconv, 1dconv, Nearest neighbor upsampling for local conditioning(Implemented)
- ema(Not implemented)