keras backend theano python wavenet
basveeling/wavenet
Once the first model checkpoint is created, you can start sampling. The latest model checkpoint will be retrieved and used to sample. We can however trade computation cost with accuracy and fidility by lowering the sampling rate, amount of stacks and the amount of channels per layer. For a downsized model (4000hz vs 16000 sampling rate, 16 filters v/s 256, 2 stacks vs??