WaveNet: A generative model for raw audio

Created on 2023-02-28T07:22:12-06:00

Return to the Index

This card pertains to a resource available on the internet.

This card can also be read via Gemini.

Synthsizes audio with a very dense convolution network.

(extremely rough approximation in ascii)

                 |
     /     /     |
  /  |  /  |  /  |
 /| /| /| /| /| /|
..................

A large amount of connections (though they get more sparse as it goes) creates individual output samples given the previous inputs of the input window.

This paper has been used in a lot of neural synthesizers. The results are very, very good--but the processing speed is fucking awful.