I just found the paper A Neural Algorithm of Artistic Style which seems to be very cool.

It describes a way to combine two images: Use the content of image A and the style of image B to create a new image. This is done with a CNN (e.g. VGG).

As may be seen on the following page it works very well: https://deepart.io/latest/

I saw that people use CNNs for audio classification (audio -> spectrogram -> CNN -> classification). Does this mean that this style transfer could also be used for audio? E.g. to change a speaker. Did this already do someone? Or are there any problems and reasons why this does not work that easy for audio?

Thank you very much

  • 1
    $\begingroup$ I don't feel like doing advertisement for a commercial product here. So I'll just say that frequently see advertisement about a program that do "sound track mastering for multiple genre using machine learning". So you can have no doubt it has been done. $\endgroup$ – Apiwat Chantawibul Jan 30 '17 at 15:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.