High-resolution radio using neural networks

A method, comprising:

• training a DNN module on a first data set of synthesized audio created through a synthesized broadcasting processor and on a second data set comprising source material audio;
• receiving, at a receiver module, a broadcast signal and providing therefrom a digital audio signal output having no audio content above a broadcast cutoff frequency;
• and
• processing, at the DNN module, the digital signal output from the receiver module to reconstruct audio comprising audio content above the broadcast cutoff frequency.

An apparatus, comprising:

• an FM radio module comprising an FM receiver module;
• the FM receiver module comprising an audio signal output having no audio content above a predetermined broadcast cutoff frequency;
• a DNN module trained on a first data set of synthesized audio created through a synthesized FM broadcasting processor and a second data set of source material audio;
• and
• the DNN module comprising an input receiving the audio signal output and an output comprising reconstructed audio content above the predetermined broadcast cutoff frequency.