Methods and systems for equalisation
A method of equalising a signal derived from an input device, the method comprising: receiving the signal; applying an order-statistic filter to the signal in the frequency domain to generate a statistically filtered signal; equalising the received signal based on the statistically filtered signal to generate an equalised signal.
1 . A method of equalising a signal derived from an input device, the method comprising:
receiving the signal;
applying an order-statistic filter directly to the signal in a frequency domain to generate a statistically filtered signal;
equalising the received signal based on the statistically filtered signal to generate an equalised signal.
2 . The method of claim 1 , wherein equalising the signal comprises:
setting an equalisation target of an equaliser to an inverse of the statistically filtered signal; and
equalising the signal using the equaliser.
3 . The method of claim 2 , wherein equalising the audio signal comprises:
identifying frequency bins of the statistically filtered signal having amplitudes greater than a threshold amplitude; and
removing components from frequency bins of the signal corresponding to the identified frequency bins of the statistically filtered signal.
4 . The method of claim 1 , wherein equalising the signal comprises:
setting an equalisation target of an equaliser to a ratio of the statistically filtered signal to the signal; and
equalising the signal using the equaliser.
5 . The method of claim 1 , further comprising smoothing the statistically filtered signal to generate a smoothed statistically filtered signal, wherein the equalising comprises equalising the smoothed statistically filtered signal.
6 . The method of claim 5 , wherein the smoothing is performed using a moving average filter or a Savitsky-Golay filter.
7 . The method of claim 6 , further comprising:
applying a threshold to the difference between the statistically filtered signal and the smooth statistically filtered signal to identify frequency bins in the statistically filtered signal having amplitudes exceeding the threshold;
wherein the equalising comprises removing or suppressing components of frequency bins of the signal corresponding to the identified frequency bins in the statistically filtered signal.
8 . The method of claim 1 , wherein the order-statistic filter comprises one of:
a) a median filter;
b) a percentile filter;
c) a minimum filter;
d) a trimmed mean filter; and
e) a trimmed minimum filter.
9 . The method of claim 8 , wherein the median filter is a recursive median filter or wherein the percentile filter is a recursive percentile filter.
10 . The method of claim 1 , wherein the order-statistic filter is applied to the entire signal.
11 . The method of claim 1 , wherein the order-statistic filter is applied to a subset of frequency bins of the signal.
12 . The method of claim 11 , wherein a frequency range of the subset of frequency bins spans a frequency of an artefact in the signal.
13 . The method of claim 1 , further comprising:
determining a discrete Fourier transform of the signal;
wherein applying an order-statistic filter to the signal in the frequency domain comprises applying the order-statistic filter to the discrete Fourier transform of the signal.
14 . The method of claim 13 , wherein equalising the received signal comprises applying weights to frequency bins of the discrete Fourier transform.
15 . The method of claim 1 , wherein the steps of applying the order-statistic filter and/or equalising the received signal are implemented using a trained neural network.
16 . The method of claim 1 , wherein one of the following applies:
a) the input device comprises a microphone and the signal comprises an audio signal;
b) the input device comprises a photodetector and the signal represents light incident the photodetector;
c) the input device comprises an accelerometer and the signal comprises an accelerometer signal.
17 . A method of equalising a signal, comprising:
determining a discrete Fourier transform for the signal;
determine a median value for each frequency bin of the discrete Fourier transform;
smoothing the median values over a frequency range of the discrete Fourier transform;
inverting the smoothed median values; and
applying a weight to each frequency bin of the discrete Fourier transform based on a corresponding one of the inverted smooth median values.
18 . The method of claim 17 , wherein the discrete Fourier transform is a fast Fourier transform (FFT).
19 . A system for equalising a signal derived from an input device, the system comprising:
an input for receiving the signal;
one or more processors configured to:
apply an order-statistic filter directly to the signal in a frequency domain to generate a statistically filtered audio signal;
equalise the signal based on the statistically filtered signal to generate an equalised signal.
20 . An electronic device comprising the system of claim 19 .
21 . The electronic device of claim 20 , wherein the device comprises one of a smartphone, a tablet, a laptop computer, a games console, a home control system, a home entertainment system, an in-vehicle entertainment system, and a domestic appliance.
22 . A non-transitory computer readable storage medium having computer-executable instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform a method of equalising a signal derived from an input device, the method comprising:
receiving the signal;
applying an order-statistic filter directly to the signal in a frequency domain to generate a statistically filtered signal; and
equalising the received signal based on the statistically filtered signal to generate an equalised signal.