Detection and suppression of howl speech signal
A system suppresses howl in a device including microphones and speakers, for example, an artificial reality headset. A speaker of the device presents audio content. The audio content presented by the speaker is received by a microphone of the device thereby creating a howl in certain situations. The system detects the presence of the howl in a region of the audio content using an adaptive notch filter. The system suppresses the howl by reducing gain of one or more frequencies of the audio content. The system may detect presence of the howl by monitoring flatness of the signal. The system may detect presence of the howl based on tonality detection based on linear prediction.
1 . A computer-implemented method for suppressing howl in a device comprising one or more microphones and one or more speakers, the computer-implemented method comprising:
presenting audio content via a speaker of the device;
receiving by a microphone of the device, the audio content presented via the speaker;
detecting presence of a howl in a region of the audio content using linear prediction to determine a location of an adaptive notch filter; and
suppressing the howl in the region of the audio content by reducing gain of one or more frequencies of the howl in the region of the audio content corresponding to the determined location of the adaptive notch filter.
2 . The computer-implemented method of claim 1 , wherein detecting presence of the howl comprises:
monitoring flatness of signal; and
responsive to detecting more than a threshold flatness of signal, determining that howl is present.
3 . The computer-implemented method of claim 1 , wherein the adaptive notch filter is one of a plurality of adaptive notch filters, wherein the audio content comprises signals of a plurality of frequency ranges, each adaptive notch filter from the plurality of adaptive notch filters associated with a frequency range from the plurality of frequency ranges, wherein the adaptive notch filter is associated with a particular frequency range, wherein suppressing the howl comprises:
reducing gain of frequencies of the particular frequency range.
4 . The computer-implemented method of claim 1 , wherein suppressing the howl further comprises:
responsive to reducing the gain of the one or more frequencies of the audio content, keeping the gain low for at least a threshold amount of time, thereby eliminating the howl.
5 . The computer-implemented method of claim 4 , further comprising: responsive to eliminating the howl, increasing gain for the one or more frequencies.
6 . The computer-implemented method of claim 1 , wherein the adaptive notch filter has two poles and two zeros.
7 . The computer-implemented method of claim 1 , wherein detecting presence of the howl comprises tonality detection based on the linear prediction.
8 . The computer-implemented method of claim 1 , wherein the device is an artificial reality headset.
9 . A non-transitory computer-readable storage medium storing instructions for suppressing howl in a device comprising one or more microphones and one or more speakers, the instructions when executed by a computer processor, cause the computer processor to perform steps comprising:
presenting audio content via a speaker of the device;
receiving by a microphone of the device, the audio content presented via the speaker;
detecting presence of a howl in a region of the audio content using linear prediction to determine a location of an adaptive notch filter; and
suppressing the howl in the region of the audio content by reducing gain of one or more frequencies of the howl in the region of the audio content corresponding to the determined location of the adaptive notch filter.
10 . The non-transitory computer-readable storage medium of claim 9 , wherein instructions for detecting presence of the howl comprise instructions that cause the computer processor to perform steps comprising:
monitoring flatness of signal; and
responsive to detecting more than a threshold flatness of signal, determining that howl is present.
11 . The non-transitory computer-readable storage medium of claim 9 , wherein the adaptive notch filter is one of a plurality of adaptive notch filters, wherein the audio content comprises signals of a plurality of frequency ranges, each adaptive notch filter from the plurality of adaptive notch filters associated with a frequency range from the plurality of frequency ranges, wherein the adaptive notch filter is associated with a particular frequency range, wherein instructions for suppressing the howl comprise instructions that cause the computer processor to perform steps comprising:
reducing gain of frequencies of the particular frequency range.
12 . The non-transitory computer-readable storage medium of claim 9 , wherein instructions for suppressing the howl comprise instructions that cause the computer processor to perform steps comprising:
responsive to reducing the gain of the one or more frequencies of the audio content, keeping the gain low for at least a threshold amount of time, thereby eliminating the howl.
13 . The non-transitory computer-readable storage medium of claim 12 , wherein the instructions further cause the computer processor to:
responsive to eliminating the howl, increasing gain for the one or more frequencies.
14 . The non-transitory computer-readable storage medium of claim 9 , wherein detecting presence of the howl comprises tonality detection based on the linear prediction.
15 . A device comprising:
one or more speakers configured to present audio content via a speaker of the device;
one or more microphones configured to receive the audio content presented via the speaker; and
a non-transitory computer-readable storage medium storing instructions for suppressing howl in a device comprising one or more microphones and one or more speakers, the instructions when executed by a computer processor, cause the computer processor to perform steps comprising:
detecting presence of a howl in a region of the audio content using linear prediction to determine a location of an adaptive notch filter; and
suppressing the howl in the region of the audio content by reducing gain of one or more frequencies of the howl in the region of the audio content corresponding to the determined location of the adaptive notch filter.
16 . The device of claim 15 , wherein instructions for detecting presence of the howl comprise instructions that cause the device to perform steps comprising:
monitoring flatness of signal; and
responsive to detecting more than a threshold flatness of signal, determining that howl is present.
17 . The device of claim 15 , wherein the adaptive notch filter is one of a plurality of adaptive notch filters, wherein the audio content comprises signals of a plurality of frequency ranges, each adaptive notch filter from the plurality of adaptive notch filters associated with a frequency range from the plurality of frequency ranges, wherein the adaptive notch filter is associated with a particular frequency range, wherein instructions for suppressing the howl comprise instructions that cause the computer processor to perform steps comprising:
reducing gain of frequencies of the particular frequency range.
18 . The device of claim 15 , wherein instructions for suppressing the howl comprise instructions that cause the computer processor to perform steps comprising:
responsive to reducing the gain of the one or more frequencies of the audio content, keeping the gain low for at least a threshold amount of time, thereby eliminating the howl.
19 . The device of claim 18 , wherein the instructions further cause the computer processor to:
responsive to eliminating the howl, increasing gain for the one or more frequencies.
20 . The device of claim 15 , wherein detecting presence of the howl comprises tonality detection based on the linear prediction.