Method and apparatus for masking wind noise

摘要

Wind and other noise is suppressed in a signal by adaptively changing characteristics of a filter. The filter characteristics are changed in response to the noise content of the signal over time using a history of noise content. Filter characteristics are changed according to a plurality of reference filters, the characteristics of which are chosen to optimally attenuate or amplify signals in a range of frequencies.

主权项

1. A method of suppressing wind noise in an audio signal obtained from a microphone, the method comprising:
receiving audio frequency signals at a low pass filter, the audio frequency signals being received by the low pass filter from a microphone, said audio frequency signals comprising at least one of: noise, speech and a combination of noise and speech; generating a wind noise probability by calculating a smoothened power ratio based on a ratio of signals output from the low pass filter to the audio frequency signals received from the microphone, wherein the smoothened power ratio represents a wind noise probability; selecting a reference filter from a plurality of reference filters based on the wind noise probability; selecting a cogent frequency of an adaptive filter coupled to the microphone and changing an attenuation of the adaptive filter at the selected cogent frequency, the adaptive filter cogent frequency being selected from a plurality of frequencies using at least one characteristic of the selected reference filter, the selecting of a cogent frequency of the adaptive filter and the changing of the attenuation of the adaptive filter being made responsive to evaluation of the wind noise probability, thereby changing a pass band characteristic of the adaptive filter; generating and outputting different filter coefficients from the adaptive filter responsive to different wind noise probabilities; and multiplying digital representations of audio frequency signals from the microphone by filter coefficients of the adaptive filter to produce a noise-reduced output signal; wherein the steps of: generating a wind noise probability, selecting a reference filter, selecting a cogent frequency of an adaptive filter, changing an attenuation of an adaptive filter, generating and outputting filter coefficients and multiplying digital representations of audio frequency signals from the microphone by coefficients of the adaptive filter, are performed continuously to continuously change a shape of the adaptive filter's pass band responsive to changing characteristics of wind noise in the audio frequency signals.