Device impairment detection

• US 9,779,759 B2
• Filed: 09/24/2015
• Issued: 10/03/2017
• Est. Priority Date: 09/17/2015
• Status: Active Grant

• Current Assignee: Sonos, Inc.
• Original Assignee: Sonos, Inc.
• Inventors: Hartung, Klaus, Bright, Gregory
• Primary Examiner(s): Teshale, Akelaw
• Application Number:
  US14/864,533
• Publication Number:
  US 20170084294A1
• Time in Patent Office: 740 Days
• Field of Search: 381 56, 381 58, 381111, 381 17, 381 57, 381 60
• CPC Class Codes: G06N 3/02 Neural networks
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: G10L 25/30 using neural networks
• CPC Class Codes: H04R 2227/005 Audio distribution systems for home, i.e. multi-room use
• CPC Class Codes: H04R 27/00 Public address systems circuits for preventing acoustic reaction H04R3/02; circuits for distributing signals to loudspeakers H04R3/12; monitoring or testing arrangements for public address systems H04R29/007; amplifiers H03F
• CPC Class Codes: H04R 29/007 for public address systems public address systems per se H04R27/00
• CPC Class Codes: H04R 3/04 for correcting frequency response
• CPC Class Codes: H04S 7/301 Automatic calibration of stereophonic sound system, e.g. with test microphone

Abstract

Examples described herein involve detecting known impairments or other known conditions using a neural network. An example implementation involves receiving data indicating a response of a playback device as captured by a microphone. The implementation also involves determining an input vector by projecting a response vector that represents the response of the playback device onto a principle component matrix representing variance caused by one or more known impairments. The implementation further involves providing the determined input vector to a neural network that includes an output layer comprising neurons that correspond to respective known impairments. The implementation involves detecting that the input vector caused one or more neurons of the neural network to fire such that the neural network indicates that a particular known impairment is affecting the microphone and/or the playback device and adjusting operation of the playback device and/or the microphone to offset the particular known impairment.

Claims

1. A non-transitory computer-readable medium having stored therein instructions executable by one or more processors to cause a control device to perform operations comprising:

receiving data indicating a detected response of one or more playback devices captured by a microphone;

determining a calibration profile of the one or more playback devices based on the data, wherein the calibration profile comprises an output frequency response of the one or more playback devices;

determining an input vector by projecting a response vector that represents the detected response of the one or more playback devices onto a principle component matrix representing variance caused by one or more calibration impairments;

providing the determined input vector to a neural network that includes an output layer comprising neurons that correspond to respective calibration impairments;

detecting that the input vector caused one or more neurons of the neural network to fire such that the neural network indicates that one or more particular calibration impairments are affecting the microphone; and

adjusting the output frequency response of the one or more playback devices to offset the one or more particular calibration impairments.

2. The non-transitory computer-readable medium of claim 1, wherein adjusting the output frequency response of the one or more playback devices to offset the one or more particular calibration impairments comprises sending a message that causes the one or more playback devices to abort calibration.

3. The non-transitory computer-readable medium of claim 1, wherein adjusting the output frequency response of the one or more playback devices to offset the one or more particular calibration impairments displaying a prompt to remove the calibration impairments from the microphone.

4. The non-transitory computer-readable medium of claim 1, wherein the input vector represents a response of the microphone in a given environment, and wherein adjusting the output frequency response of the one or more playback devices to offset the one or more particular calibration impairments comprises:

applying one or more respective correction curves corresponding to the one or more particular calibration impairments to the output frequency response to offset acoustic characteristics of the given environment to calibrate the one or more playback devices to a calibration equalization.

5. The non-transitory computer-readable medium of claim 1, wherein the neural network includes at least one first neuron corresponding to an absence of calibration impairment and one or more second neurons corresponding to respective calibration impairments, and wherein detecting that the input vector caused one or more neurons of the neural network to fire comprises:

determining that the determined input vector did not cause the at least one first neuron to fire; and

determining that the input vector caused at least one of the one or more second neurons to fire.

6. The non-transitory computer-readable medium of claim 1, wherein detecting that the input vector caused one or more neurons of the neural network to fire comprises determining that output value of neuron transfer functions corresponding to respective neurons exceed a firing threshold.

7. The non-transitory computer-readable medium of claim 1, wherein detecting that the input vector caused one or more neurons of the neural network to fire comprises detecting that the input vector caused a neuron corresponding to a particular type of recording-device case to fire.

8. The non-transitory computer-readable medium of claim 1, wherein detecting that the input vector caused one or more neurons of the neural network to fire comprises detecting that the input vector caused a neuron corresponding to a particular type of room to fire.

9. The non-transitory computer-readable medium of claim 1, wherein determining the input vector comprises:

identifying a subset of the received data indicating the response of one or more playback devices, the subset corresponding to frequencies affected by the calibration impairments; and

determining a response vector that includes components representing the identified subset.

10. The non-transitory computer-readable medium of claim 9, wherein determining the input vector that includes components representing the identified subset comprises:

converting the received data indicating the response of one or more playback devices from linear spacing to logarithmic spacing; and

determining a response vector that includes log-spaced components representing the identified subset.

11. The non-transitory computer-readable medium of claim 1, wherein receiving data indicating the detected response of the one or more playback devices comprises receiving data indicating a detected response of one or more first playback devices of a first type in a given environment, and

wherein the operations further comprise:

determining a difference between the detected response of the one or more first playback devices and a detected response corresponding to a second type of playback device, wherein the neural network was trained with a second playback device of the second type; and

determining the response vector by determining a product of an inverse of the determined difference and the detected response of the one or more first playback devices so as to offset the different types of playback devices.

12. A playback device comprising:

one or more processors;

a tangible data storage having stored therein instructions executable by the one or more processors to cause the playback device to perform operations comprising;

receiving data indicating a detected response of the playback device in a given environment which was captured with a recording device;

determining a calibration profile of the playback device based on the data, wherein the calibration profile comprises an output frequency response of the playback device;

determining an input vector by projecting a response vector that represents the detected response of the playback device onto a principle component matrix representing variance caused by one or more conditions;

providing the determined input vector to a neural network that includes an output layer comprising neurons that correspond to respective conditions;

detecting that the input vector caused one or more neurons of the neural network to fire such that the neural network indicates that one or more particular conditions are affecting the recording device; and

adjusting the output frequency response of the playback device to offset the one or more particular conditions.

13. The playback device of claim 12, wherein adjusting the output frequency response of the playback device to offset the one or more particular conditions comprises:

interrupting calibration of the playback device;

while calibration is interrupted, causing a control device to display a prompt to remove the conditions; and

upon receiving an indication that the conditions are removed, continuing calibration of the playback device.

14. The playback device of claim 12, wherein adjusting the output frequency response of the playback device to offset the one or more particular conditions comprises:

applying one or more respective correction curves corresponding to the one or more particular conditions to the output frequency response to offset acoustic characteristics of the given environment to calibrate the playback device to a calibration equalization.

15. The playback device of claim 12, wherein the neural network includes at least one first neuron corresponding to an absence of calibration impairment and one or more second neurons corresponding to respective calibration impairment, and wherein detecting that the input vector caused one or more neurons of the neural network to fire comprises:

determining that the determined input vector did not cause the at least one first neuron to fire; and

determining that the input vector caused at least one of the one or more second neurons to fire.

16. A method comprising:

receiving, by a computing device, data indicating a detected response of one or more playback devices captured by a microphone;

determining, by the computing device, a calibration profile of the one or more playback devices based on the data, wherein the calibration profile comprises an output frequency response of the one or more playback devices;

determining, by the computing device, an input vector by projecting a response vector that represents the detected response of the one or more playback devices onto a principle component matrix representing variance caused by one or more calibration impairments;

providing, by the computing device, the determined input vector to a neural network that includes an output layer comprising neurons that correspond to respective calibration impairments;

detecting, by the computing device, that the input vector caused one or more neurons of the neural network to fire such that the neural network indicates that one or more particular calibration impairments are affecting the microphone; and

adjusting, by the computing device, the output frequency response of the one or more playback devices to offset the one or more particular calibration impairments.

17. The method of claim 16, wherein receiving data indicating the detected response of one or more playback devices captured by the microphone comprises:

causing a microphone of the computing device to detect a calibration sound emitted by the one or more playback devices, the calibration sound covering a calibration frequency range.

18. The method of claim 16, wherein adjusting the output frequency response of the one or more playback devices of the calibration profile of the one or more playback devices to offset the one or more particular calibration impairments comprises causing a control device to display a prompt to remove the calibration impairments from the microphone.

19. The method of claim 16, wherein the input vector represents a response of the microphone in a given environment, and wherein adjusting the output frequency response of the one or more playback devices to offset the one or more particular calibration impairments comprises:

applying one or more respective correction curves corresponding to the one or more particular calibration impairments to the output frequency response to offset acoustic characteristics of the given environment to calibrate the one or more playback devices to a calibration equalization.

20. The method of claim 16, wherein receiving data indicating the detected response of the one or more playback devices comprises receiving data indicating a response of one or more first playback devices of a first type in a given environment, and wherein the method further comprises:

determining a difference between the detected response of the one or more first playback devices and a response corresponding to a second type of playback device, wherein the neural network was trained with a second playback device of the second type; and

determining the response vector by determining a product of an inverse of the determined difference and the detected response of the one or more first playback devices so as to offset the different types of playback devices.