System for elimination of acoustic feedback
A method and system is provided for eliminating acoustical feedback in a system. The method determines a parameter for at least one notch filter, adjusting the notch filter based on the parameter, processing the digital signals through the notch filter, testing at the effect of the notch filter in the system, and removing the notch filter if the notch filter is not effective. Also disclosed is a method and system of selecting candidate frequencies which might be feedback, as opposed to other wanted sound frequencies. The selection method sampling the digital signals, converting the time domain digital signal samples by a fast Fourier transform algorithm into the frequency domain, using a ballistics approach to find prominences in the frequency spectrum, and testing the sizes of the prominences.
1. A method of reducing acoustic feedback in a system comprising:
sampling a sound signal at predetermined intervals to iteratively create a set of sampled sound signals;
transforming with a processor, each one of the set of sampled sound signals to a frequency spectrum comprising a plurality of frequency magnitude bins, each of the frequency magnitude bins comprising a bin value indicative of a magnitude of a frequency of the sound signal over a predetermined frequency range;
comparing an old bin. value in a first sampled sound signal. included in the set of sampled sound signals with a new bin value in a second sampled sound signal included in the set of sampled sound signals; and
setting the old bin value in the first sampled sound signal equal to the new bin value in the second sampled sound signal in response to the new bin value in the second sampled sound signal being less than the old bin value in the first sampled sound signal;
increasing the old bin value in the first sampled sound signal as a function of the old bin value in the first sampled sound signal, the new bin value in the second sampled sound signal and a determined filter coefficient in response to the new bin value in the second sampled sound signal being greater than the old bin value in the first sampled sound signal;
selecting a first frequency magnitude bin to be a candidate frequency; and
setting a notch filter to the candidate frequency at an attenuation depth and a bandwidth to suppress feedback.
2. The method of claim 1 , where increasing the first value includes calculating the filter coefficient as a function of a threshold, a time constant, and a frame sample.
3. The method of claim 1 , where selecting the first frequency magnitude bin includes determining that the first value is at least a predetermined magnitude.
4. The method of claim 3 , where the filter coefficient includes an adjustable time constant.
5. The method of claim 4 , where the filter coefficient includes monitoring the first frequency magnitude bin over a determined number of frames based on the adjustable time constant.
6. The method of claim 5 , where monitoring further includes adjusting the adjustable time constant based on a frequency of sound being represented in the first frequency magnitude bin.
7. The method of claim 5 , where monitoring further includes applying a longer time constant to the first frequency magnitude bin in response to the first frequency magnitude bin being representative of a lower sound frequency, and
applying a shorter time constant to the first frequency magnitude bin in response to the first frequency magnitude bin being representative of a higher sound frequency.
8. The method of claim 5 , further includes determining the adjustable time constant as the time taken to reach 6 dB below a threshold.
9. The method of claim 5 , further includes limiting the adjustable time constant to a range between 200 milliseconds to 2 seconds.
10. The method of claim 5 , where monitoring further includes comparing an amplitude of the first value to a relative threshold of the buffered acoustic signal and an absolute threshold.
11. The method of claim 10 , where comparing includes calculating with a processor, the relative threshold as a function of energy level of the buffered acoustic signal and a predetermined relative multiplier.
12. The method of claim 10 , where the absolute threshold is a determined decibel level.