In this paper we present an example-driven algorithm that allows the recovery of wide regions of lost spectral compo- nents in bandlimited signals. We present a generative spectral model which allows the extraction of salient information from audio snippets, and then apply this information to enhance the bandwidth of bandlimited signals.
Learn More