Single-channel, binaural and multi-channel dereverberation

A method, in a multimedia signal processing system, that allows a user to adjust suppression of reverberation or noise from a digital signal that represents a sound signal, comprising:

• analyzing the digital signal by determining a plurality of time frequency frames of the digital signal;
• deriving a first estimation of reverberation or noise from one of the frames in a first time instant;
• deriving a first signal-to-noise ratio or signal-to-reverberation ratio;
• deriving a first spectral gain from the first estimation;
• automatically determining a first exponent based upon the first signal-to-noise ratio or signal-to-reverberation ratio;
• deriving a modified first spectral gain based on the first spectral gain raised to a power related to the first exponent;
• wherein the modified first spectral gain is applied to the one frame in the first time instant;
• deriving a second estimation of reverberation or noise from one of the frames in a second time instant;
• deriving a second signal-to-noise ratio or signal-to-reverberation ratio;
• deriving a second spectral gain from the second estimation;
• automatically determining a second exponent based upon the second signal-to-noise ratio or signal-to-reverberation ratio;
• deriving a modified second spectral gain based on the second spectral gain raised to a power related to the second exponent;
• wherein the modified second spectral gain is applied to the one frame in the second time instant;
• outputting a signal that involves processing the frame in the first time instant utilizing the first modified spectral gain;
• and outputting a signal that involves processing the frame in the second time instant utilizing the second modified spectral gain;
• wherein the first exponent and the second exponent are different from one another, and wherein the second time instant is subsequent to the first time instant.