Recent research in machine learning has focused on breaking audio spectrograms into separate sources of sound using latent variable decompositions. These methods require that the ...
Wiener filtering is one of the most widely used methods in audio source separation. It is often applied on time-frequency representations of signals, such as the short-time Fourier...
Jonathan Le Roux, Emmanuel Vincent, Yuu Mizuno, Hi...
Various methods have recently appeared to transform foreign-accented speech into its native-accented counterpart. Evaluation of these accent conversion methods requires extensive l...