Frequency domain noise cancellation with a desired null based acoustic devices, systems, and methods

A system for signal extraction in a frequency domain, comprising:

• an FFT component, the FFT component receives a first signal which contains mostly undesired audio, substantially void of desired audio, the FFT component decomposes the first signal into at least two reference frequency components;
• at least two adaptive filters, each of the at least two reference frequency components is filtered by a corresponding adaptive filter of the at least two adaptive filters to provide filtered reference frequency components;
• an IFFT component, each of the filtered reference frequency components is input into the IFFT component and recombined into a filtered reference signal,
• an adder, the filtered reference signal is input into the adder and a second signal which contains delayed desired audio and undesired audio is input into the adder, the adder subtracts the filtered reference signal from the second signal to output desired audio;
• an FFT component, the FFT component is coupled to an output of the adder and decomposes the output of the adder into at least two error signal frequency components, each error signal frequency component, of the at least two error signal frequency components, is input into a corresponding adaptive filter of the at least two adaptive filters and each error signal frequency component is used to adapt a respective adaptive filter;
• and
• an inhibit component, the inhibit component includes inhibit logic and performs inhibition in the frequency domain by controlling the at least two adaptive filters.

A system for signal extraction in a frequency domain, comprising:

• an first analysis bank, the first analysis bank receives a first signal which contains mostly undesired audio, substantially void of desired audio, the first analysis bank decomposes the first signal into at least two reference sub-band components;
• at least two adaptive filters, each of the at least two reference sub-band components is filtered by a corresponding adaptive filter of the at least two adaptive filters to provide filtered reference sub-band components;
• a second analysis bank, the second analysis bank receives a delayed second signal which contains desired audio and undesired audio, the second analysis bank decomposes the delayed second signal into at least two primary sub-band components;
• at least two adders, each of the at least two primary sub-band components is input into a corresponding adder of the at least two adders, wherein a corresponding filtered reference sub-band component is subtracted from a corresponding primary sub-band component in the corresponding adder, an output from the corresponding adder is fed back to the corresponding adaptive filter to provide information that is used during the filtering at the corresponding adaptive filter;
• an inhibit component, the inhibit component includes inhibit logic and uses inputs from the at least two primary sub-band components and the at least two reference sub-band components to control the at least two adaptive filters;
• and
• a synthesis bank, outputs from the at least two adders are input into the synthesis bank and recombined to form a recombined signal, the recombined signal is output as desired audio.

A system for signal extraction in a frequency domain, comprising:

• a first analysis bank, the first analysis bank receives a first signal which contains mostly undesired audio, substantially void of desired audio, the first analysis bank decomposes the first signal into at least two reference frequency sub-band components;
• at least two adaptive filters, each of the at least two reference frequency sub-band components is filtered by a corresponding adaptive filter of the at least two adaptive filters to provide filtered reference frequency sub-band components;
• a synthesis bank, each of the filtered reference frequency sub-band components is input into the synthesis bank and recombined into a filtered reference signal,
• an adder, the filtered reference signal is input into the adder and a second signal which contains delayed desired audio and undesired audio is input into the adder, the adder subtracts the filtered reference signal from the second signal to output desired audio;
• a second analysis bank, the second analysis bank is coupled to an output of the adder and decomposes the output of the adder into at least two error signal frequency sub-band components, each error signal frequency sub-band component, of the at least two error signal frequency sub-band components, is input into the corresponding adaptive filter of the at least two adaptive filters and each error signal frequency sub-band component is used to adapt the corresponding adaptive filter;
• and
• an inhibit component, the inhibit component includes inhibit logic and performs inhibition in sub-bands of the frequency domain by controlling the at least two adaptive filters.

A frequency domain signal extraction method, comprising:

• receiving a reference signal, the reference signal contains mostly undesired audio, substantially void of desired audio;
• decomposing the reference signal into at least two reference frequency components;
• receiving a delayed signal, the delayed signal contains desired audio and undesired audio;
• decomposing the delayed signal into at least two primary frequency components;
• filtering the at least two reference frequency components with at least two adaptive filters to form at least two filtered reference frequency components;
• inhibiting the filtering of the at least two reference frequency components based on the at least two reference frequency components and the at least two primary frequency components;
• subtracting the at least two filtered reference frequency components from the at least two primary frequency components to form at least two frequency components that contain desired audio;
• adapting the filtering with the at least two frequency components;
• and
• recombining the at least two frequency components to form a signal that contains desired audio

A frequency domain signal extraction method, comprising:

• receiving a reference signal, the reference signal contains mostly undesired audio, substantially void of desired audio;
• decomposing the reference signal into at least two reference frequency components;
• filtering the at least two reference frequency components with at least two adaptive filters to form at least two filtered reference frequency components;
• recombining the filtered reference frequency components in an IFFT component, to produce a filtered reference signal;
• inputting a delayed signal at an adder, the delayed signal contains desired audio and undesired audio;
• subtracting the filtered reference signal from the delayed signal to form an output signal containing desired audio;
• decomposing the output signal into at least two frequency components;
• adapting the filtering with the at least two frequency components;
• and
• inhibiting the filtering intermittently with the at least two adaptive filters to prevent cancellation of desired audio.