Method for processing speech signals using an ensemble of speech enhancement procedures

摘要

A method processes an acoustic signal that is a mixture of a target signal and interfering signals by first enhancing the acoustic signal by a set of enhancement procedures to produce a set of initial enhanced signals. Then, an ensemble learning procedure is applied to the acoustic signal and the set of initial enhancement signals to produce features of the acoustic signal.

主权项

1. A method for processing an acoustic signal, wherein the acoustic signal is a mixture of a target signal and interfering signals, comprising the steps of:
enhancing the acoustic signal by a set of enhancement procedures to produce a set of initial enhanced signals; obtaining, for each initial enhanced signal an input mask, which represents an effect of the corresponding enhancement procedure on the acoustic signal, by comparing the initial enhanced signal to the acoustic signal: determining features of the target signal using a combination of the input masks, wherein parameters of the combination are trained on data processed using the set of enhancement procedures; and synthesizing the target signal from the acoustic signal using the features, wherein the steps are performed in a processor.