arXiv:1801.01589 — the Stanford paper introducing the CNN-based neural style transfer method for audio spectrograms described in the parent node. https://arxiv.org/abs/1801.01589

Google

A CNN-based neural style transfer method developed at Stanford University that represents music as audio spectrograms and applies convolutional-neural-network style-transfer techniques to them, transferring style while preserving content. It is regarded as one of the strongest introductory reproduction projects in the music style transfer area. Introducing paper: https://arxiv.org/abs/1801.01589. Reference implementation: https://github.com/alishdipani/Neural-Style-Transfer-Audio.

Learn Before

Related