CPC G10L 21/038 (2013.01) [G10L 19/0208 (2013.01); G10L 25/78 (2013.01); H03G 3/20 (2013.01); G10L 19/26 (2013.01)] |
AS A RESULT OF REEXAMINATION, IT HAS BEEN DETERMINED THAT: |
The patentability of claims 1-18 is confirmed. |
1. A method comprising:
generating an excitation signal from an audio signal, wherein in the audio signal comprises a plurality of frequency components;
extracting a feature vector from the audio signal, wherein the feature vector comprises at least one frequency domain component feature and at least one time domain component feature;
determining at least one spectral shape parameter from the feature vector, wherein the at least one spectral shape parameter corresponds to a sub band signal comprising frequency components which belong to a further plurality of frequency components; and
generating the sub band signal by filtering the excitation signal through a filter bank and weighting the filtered excitation signal with the at least one spectral shape parameter,
wherein the spectral shape parameter is a sub band energy level value and the sub band energy level value is attenuated when the power of the audio signal approaches an estimate of the level of noise in the audio signal.
|