Annoyance noise suppression

1. A personal audio system, comprising:

a class table configured to store processing parameters respectively associated with a plurality of annoyance noise classes;

a controller configured to:

identify an annoyance noise class of the annoyance noise included in an ambient audio stream; and

retrieve, from the class table one or more processing parameters associated with the identified annoyance noise class;

a processor configured to process the ambient audio stream according to the one or more processing parameters associated with the identified annoyance noise class to provide a personal audio stream, the processor further comprising:

a pitch tracker configured to identify a fundamental frequency of the annoyance noise; and

a filter bank including a band reject filter tuned to the fundamental frequency;

a voice activity detector configured to detect a voice activity in the ambient audio stream,

wherein when the voice activity detector detects the voice activity, the filter bank implements a first filter function, wherein an attenuation of the first filter function is between 10 and 15 dB, and

wherein when the voice activity detector does not detect the voice activity, the filter bank implements a second filter function that differs from the first filter function, wherein an attenuation of the second filter function is between 20 and 25 dB; and

a speaker that is configured to output the personal audio stream corresponding to the ambient audio stream that has been processed by one of the first filter function and the second filter function, according to the voice activity detected by the voice activity detector.

2. The personal audio system of claim 1 , wherein

the one or more processing parameters associated with the identified annoyance noise class includes a specified frequency range, and

the pitch tracker is constrained to identify a frequency within the specified frequency range.

3. The personal audio system of claim 1 , wherein

the one or more processing parameters associated with the identified annoyance noise class includes a specified Q value, and

the band reject filter tuned to the fundamental frequency is configured to provide the specified Q value.

4. The personal audio system of claim 1 , wherein

the one or more processing parameters associated with the identified annoyance noise class includes a specified bandwidth, and

the band reject filter tuned to the fundamental frequency is configured to provide the specified bandwidth.

5. The personal audio system of claim 1 , wherein

the one or more processing parameters associated with the identified annoyance noise class includes a number of harmonics N, where N is a positive integer, and

the at least one band reject filter comprises N band reject filters tuned to N different harmonics of the fundamental frequency.

6. The personal audio system of claim 1 , wherein

the class table stores a respective profile for each of the plurality of annoyance noise classes, and

the controller is further configured to identify the annoyance noise class of the annoyance noise included in the ambient audio stream at least in part by:

determine a profile of the ambient audio stream;

compare the profile of the ambient audio stream with the profiles stored in the class table; and

identify the annoyance noise class having a profile that most closely matches the profile of the ambient audio stream.

7. The personal audio system of claim 1 , wherein the controller is configured to identify the annoyance noise class of the annoyance noise included in the ambient audio stream at least in part by:

determine a profile of the ambient audio stream;

send a query including the profile of the ambient audio stream and context information to a noise database; and

receive, from the noise database, information designating the identified annoyance noise class.

8. The personal audio system of claim 1 , wherein the controller is further configured to identify the annoyance class of the annoyance noise included in the ambient audio stream at least in part on a context of a user.

9. The personal audio system of claim 8 , wherein the context includes one or more of a physical location, activity of the user, a date, and/or time of day.

10. The personal audio system of claim 8 , wherein information associated with the context of the user is used to query to a sound database, wherein the sound database is configured to select one or more candidate annoyance noise classes as the identified annoyance noise class.

11. The personal audio system of claim 1 , wherein the first filter function suppresses the fundamental component and harmonic components of the annoyance noise, and the second filter function is a compromise between annoyance noise suppression and speech recognition.

12. The personal audio system of claim 1 , wherein the first filter function has a lower attenuation and a narrower bandwidth than the second filter function.

13. The personal audio system of claim 1 , wherein the one or more processing parameters associated with the annoyance noise class include an anticipated frequency modulation scheme for the annoyance noise class.

14. The personal audio system of claim 1 , wherein the one or more processing parameters associated with the annoyance noise class include a maximum expected rate of change of the fundamental frequency for the annoyance noise class.

15. The personal audio system of claim 1 , wherein the one or more processing parameters associated with the annoyance noise class include filter parameters for a number of harmonics to be filtered differently by the first filter function and the second filter function.

16. The personal audio system of claim 1 , wherein the controller identifies the annoyance noise class using at least two of an overall loudness level, a normalized loudness of a first set of frequency bands, an absolute loudness of a second set of frequency bands, a spectral envelop shape, a rising pitch spectrographic feature, a falling pitch spectrographic feature, a presence of a dominant narrow-band sound, a presence of odd harmonics, an absence of off harmonics, a presence of even harmonics, an absence of even harmonics, and a low frequency periodicity.

17. A method for suppressing an annoyance noise included in an ambient audio stream, comprising:

identifying an annoyance noise class of the annoyance noise included in the ambient audio stream;

retrieving, from a class table storing processing parameters respectively associated with a plurality of annoyance noise classes, wherein one or more of the processing parameters are associated with the identified annoyance noise class;

processing the ambient audio stream according to the one or more processing parameters associated with the identified annoyance noise class to generate a personal audio stream, processing the ambient audio stream further comprising:

identifying a fundamental frequency of the annoyance noise; and

filtering the ambient audio stream with a band reject filter tuned to the fundamental frequency;

detecting a voice activity in the ambient audio stream,

wherein when the voice activity is detected, the band reject filter implements a first filter function, wherein an attenuation of the first filter function is between 10 and 15 dB, and

wherein when the voice activity is not detected, the band reject filter implements a second filter function that differs from the first filter function, wherein an attenuation of the second filter function is between 20 and 25 dB; and

outputting, using a speaker, the personal audio stream corresponding to the ambient audio stream that has been processed by one of the first filter function and the second filter function, according to the voice activity having been detected.

18. The method of claim 17 , wherein

the class table stores a respective profile for each of the plurality of annoyance noise classes, and

identifying an annoyance noise class of the annoyance noise comprises:

determining a profile of the ambient audio stream;

comparing the profile of the ambient audio stream with the profiles stored in the class table; and

identifying the annoyance noise class having a profile that most closely matches the profile of the ambient audio stream.

19. The method of claim 17 , wherein identifying an annoyance noise class of the annoyance noise comprises:

determining a profile of the ambient audio stream;

sending a query including the profile of the ambient audio stream and context information to a noise database; and

receiving, from the noise database, information designating the identified annoyance noise class.

20. A computer program product for suppressing an annoyance noise included in an ambient audio stream, the computer program product being embodied in a tangible non-transitory computer readable storage medium and comprising computer instructions for:

identifying an annoyance noise class of the annoyance noise included in the ambient audio stream;

retrieving, from a class table storing processing parameters respectively associated with a plurality of annoyance noise classes, wherein one or more of the processing parameters are associated with the identified annoyance noise class;

processing the ambient audio stream according to the one or more processing parameters associated with the identified annoyance noise class to generate a personal audio stream, processing the ambient audio stream further comprising:

identifying a fundamental frequency of the annoyance noise; and

filtering the ambient audio stream with a band reject filter tuned to the fundamental frequency;

detecting a voice activity in the ambient audio stream,

wherein when the voice activity is detected, the band reject filter implements a first filter function, wherein an attenuation of the first filter function is between 10 and 15 dB, and

wherein when the voice activity is not detected, the band reject filter implements a second filter function that differs from the first filter function, wherein an attenuation of the second filter function is between 20 and 25 dB; and

controlling a speaker to output the personal audio stream corresponding to the ambient audio stream that has been processed by one of the first filter function and the second filter function, according to the voice activity having been detected.