search

rpatrik96 / pytorch-lightning-wavenet

An implementation of WaveNet using PyTorch & PyTorch Lightning
MIT License
13 stars 3 forks source link

How can this run on audio? #1

Open turian opened 4 years ago

turian commented 4 years ago

Can you explain how to run this on audio? I see that you are training on the PTB.

rpatrik96 commented 4 years ago

Hi @turian! Thanks for reaching out to me :). For using this model on audio, you would need to create your dataset and set the num_classes parameter according to that. As audio is continuous, you may need to quantize your data to convert it to a classification problem, as it was done in the original WaveNet paper (equation in section 2.2).

I hope this helps. Nonetheless, if you are still unsure, please contact me.