Device arbitration using acoustic characteristics
AssigneeAmazon Technologies, Inc.
InventorsHenry Michael D Souza, Ragini Prasad, Rick Carragher, Joseph White
Life of this patent
Estimating the term…
Abstract
Described herein is a system for device arbitration using acoustic characteristics of a physical space, such as a user's household. The system generates a matrix of inter-device attenuation factors. The inter-device attenuation factors are determined using the attenuation experienced by a first device versus the attenuation experienced by a second device. Once the matrix is generated, an attenuation vector representing the attenuation corresponding to an input audio signal is determined and compared to the matrix. Based on the comparison, the system selects a device for further processing.
Claims20
A computer-implemented method comprising:
- determining first audio data received by a first device, the first audio data corresponding to a first utterance;
- determining second audio data received by a second device, the second audio data corresponding to the first utterance;
- determining a reference inter-device attenuation factor using a first ratio and a second ratio, the first ratio between the first audio data and the second audio data, and the second ratio between a first microphone gain associated with the first device and a second microphone gain associated with the second device;
- generating a matrix using the reference inter-device attenuation factor, the matrix representing an approximate location of the first device with respect to the second device;
- receiving, from the first device, first input audio data corresponding to a second utterance originating from a user location;
- receiving, from the second device, second input audio data corresponding to the second utterance;
- calculating a first attenuation factor associated with the first input audio signal, the first attenuation factor corresponding to the first device and the user location;
- calculating a second attenuation factor associated with the second input audio signal, the second attenuation factor corresponding to the second device and the first user location;
- and
- selecting the first device from which to receive additional input audio signal using the first attenuation factor, the second attenuation factor and the matrix, the selecting based on the first attenuation factor being within a threshold of the reference inter-device attenuation factor.
A computer-implemented method comprising:
- determining a matrix representing an approximate location of a first device with respect to at least a second device, the matrix determined based at least in part on a first inter-device attenuation factor corresponding to the first device and the second device;
- determining a first attenuation factor associated with first audio data received by the first device, the first audio data corresponding to an utterance;
- determining a second attenuation factor associated with second audio data received by the second device, the second audio data corresponding to the utterance;
- and
- selecting the first device to receive additional audio data, the selecting based on the first attenuation factor, the second attenuation factor, and the matrix.
A system comprising:
- at least one processor;
- and
- at least one memory including instructions that, when executed by the at least one processor, cause the system to:
- determine a matrix representing an approximate location of a first device with respect to at least a second device, the matrix determined based at least in part on a first inter-device attenuation factor corresponding to the first device and the second device;
- determine a first attenuation factor associated with first audio data received by the first device, the first audio data corresponding to an utterance;
- determine a second attenuation factor associated with second audio data received by the second device, the second audio data corresponding to the utterance;
- and
- select the first device to receive additional audio data based on the first attenuation factor, the second attenuation factor, and the matrix.
Dependencies are inferred from the claim text.