Lightweight WaveNet in TensorFlow
Work in progress! Please star it in the mean time. Coming soon :)
The aim of this project is to provide a lightweight WaveNet implementation in Tensorflow. When I say lightweight, it means without the WAV support (no processing such as mu-law encoder and no generation). In fact, the WaveNet architecture can be applied to any kind of Time Series.
- Pip setup added (Nov, 25th)
- Loss and prediction workflows were optimized. The network is much faster now! (Nov, 20th)
- First draft pushed (Nov, 19th)