Sarroff, A. M., & Casey, M. (2014). Musical audio synthesis using autoencoding neural nets. In Joint 40th International Computer Music Conference (ICMC) and 11th Sound & Music Computing conference (SMC).
Complex-valued neural networks (CVNNs) are an emerging field of research in
neural networks due to their potential representational properties for audio, image,
and physiological signals.
Autoencoders
Espectral
http://www.cs.dartmouth.edu/~sarroff/projects/autoencoding-synthesizers/
Sarroff, A. M., & Casey, M. (2014). Musical audio synthesis using autoencoding neural nets. In Joint 40th International Computer Music Conference (ICMC) and 11th Sound & Music Computing conference (SMC).
http://www.cs.dartmouth.edu/~sarroff/papers/sarroff2014a.pdf
Repos:
https://github.com/pkmital/autoEncoderSynthesis
https://github.com/woodshop/deepAutoController
https://deepmind.com/blog/wavenet-generative-model-raw-audio/
https://arxiv.org/pdf/1609.03499.pdf