banner.jpeg
MIDS Capstone Project Fall 2020

DeciBull

DeciBull is a neural network based audio upsampling product which up-samples low quality audio to higher quality (super resolution audio) to improve user experience on online conference platforms such as Zoom, Skype etc. DeciBull is trained on more than 10GB high quality speech data and uses Generative Adversarial networks (GANs) to generate upsampled super resolution audio.   

 

Last updated:

December 10, 2020