MelSpecVAE: Variational Autoencoder for one-shot audio synthesis

One-Shot sample synthesis of MelSpecVAE trained for 360 epochs on a DJ set by artist amazondotcom: https://www.youtube.com/watch?v=kwnTEJADYDE&t=2990s

MelSpecVAE: One shot sample synthesis

Sound sample of interpolation between 2 seeds of MelSpecVAE.

Model was trained for 360 epochs on a DJ set by artist @amazondotcom: https://www.youtube.com/watch?v=kwnTEJADYDE&t=2990s

MelSpecVAE: Interpolation between 2 seeds

Interpolation between two randomly generated seeds. Training was done with my own back-catalog of music.

You can find the code for MelSpecVAE here: https://github.com/moiseshorta/MelSpecVAE

MelSpecVAE: Interpolation between 2 seeds (ex.2)

" ℌEXOℜℭℑSMOS is an uncanny link between ancient and contemporary sound technologies. Prehispanic sound artifacts from ancient México, contemporary brain-computer interfaces and hand crafted electronic instruments blend together to situate the listener in a teknoXamanistic ritual of rhythmical noise soundscapes which addresses the politics of colonization and occupation – whether it be on the dance floor or interactive immersive sound installations.
 
He's an alumni of the Red Bull Music Academy held in London, UK (2010), participated as a guest composer at ElektronMusikStudion in Stockholm, Sweden (2014 & 2017), played in Boiler Room México, and has participated in the Tijuana, Mexico City and Berlin editions of Music Makers Hacklab by CTM festival. " 

• contact: hola [at] moiseshorta.audio

⏈⏅⏂⏈⏅⏂⏈⏅⏂⏈⏅⏂⏈
•●⑀●⑀●⑀●⑀●⑀●⑀●⑀●⑀●⑀●⑀●•
⏁⏄⏇⏁⏄⏇⏁⏄⏇⏁⏄⏇⏁
▁▂▃▄▅▆▇▆▅▄▃▂▁
⏈⏅⏂⏈⏅⏂⏈⏅⏂⏈⏅⏂⏈
•●⑀●⑀●⑀●⑀●⑀●⑀●⑀●⑀●⑀●⑀●•

MelSpecVAE: Variational Autoencoder for one-shot audio synthesis

𝔥𝔢𝔵𝔬𝔯𝔠𝔦𝔰𝔪𝔬𝔰