4

I recently read Bytenet and Wavenet and I was curious why the first model is not as popular as the second. From my understanding, Bytenet can be seen as a seq2seq model where the encoder and the decoder are similar to Wavenet. Following the trends from NLP where seq2seq models seem to perform better, I find it strange that I couldn't find any paper that compares the two. Are there any drawbacks of Bytenet over Wavenet other than the computation time?

nbro
  • 39,006
  • 12
  • 98
  • 176
razvanc92
  • 1,108
  • 7
  • 18
  • 1
    Have you seen this thread https://www.reddit.com/r/MachineLearning/comments/ai4vro/d_anyone_here_use_bytenet_like_architecture_for/ ? – Brian O'Donnell Aug 28 '19 at 13:15
  • 1
    David Pollack mentions some characteristics of the two models in his master's thesis: "Musical Genre Classification of Audio" at https://edoc.hu-berlin.de/bitstream/handle/18452/20012/master_pollack_david.pdf?sequence=7 – Brian O'Donnell Aug 28 '19 at 13:32

1 Answers1

0

My conclusion is the same as yours that there doesn't seem to be any published comparison of the two models. ByteNet is computationally expensive and requires a lot of parameters. WaveNet improves on ByteNet's efficiency, as you mentioned, and I believe that is the main difference.

Brian O'Donnell
  • 1,853
  • 6
  • 20