US 12,308,018 B2 · Systems and methods for generating labeled data to facilitate configuration of network microphone devices

Abstract

Systems and methods for generating training data are described herein. Pieces of metadata captured by a plurality of networked sensor systems can be captured, where each piece of metadata is associated with a specific set of sensor data captured by one of the plurality of networked sensor systems and includes a set of characteristics for the specific set of captured sensor data. A probabilistic model can be generated based on the received metadata and simulations can be performed based upon a training corpus by generating multiple scenarios, and, for each scenario, a scenario specific version of a particular annotated sample is generated by performing a simulation using the particular annotated sample. The scenario specific versions of annotated samples from the training corpus can be stored as a training data set on the at least one network device.

Claims21

Dependent claims

A method for generating training data that simulates sound collected in various operating environments, the method comprising:

receiving sound metadata that characterizes a plurality of sets of sound data;
generating a probabilistic model based on the received sound metadata, wherein the probabilistic model comprises a plurality of probability distribution functions for sound data characteristics that characterize the plurality of sets of sound data;
performing an acoustic simulation to obtain a noised version of an annotated speech sample, wherein performing the acoustic simulation comprises:
generating a scenario that comprises a set of sound data characteristics that are drawn from the probabilistic model, and
generating the noised version of the annotated speech sample that corresponds to the generated scenario by performing the acoustic simulation using the set of sound data characteristics;
and
simulating performance of a simulated network microphone device using the noised version of the annotated speech sample.

4 more elements

A computer system comprising:

a communication interface configured to facilitate communication via at least one data network;
at least one processor;
and
at least one non-transitory computer-readable medium comprising program instructions that are executable by the at least one processor such that the computer system is configured to:
receive sound metadata that characterizes a plurality of sets of sound data;
generate a probabilistic model based on the received sound metadata, wherein the probabilistic model comprises a plurality of probability distribution functions for sound data characteristics that characterize the plurality of sets of sound data;
perform an acoustic simulation to obtain a noised version of an annotated speech sample, wherein performing the acoustic simulation comprises:
generating a scenario that comprises a set of sound data characteristics that are drawn from the probabilistic model, and
generating the noised version of the annotated speech sample that corresponds to the generated scenario by performing the acoustic simulation using the set of sound data characteristics;
and
simulate performance of a simulated network microphone device using the noised version of the annotated speech sample.

8 more elements

A non-transitory computer readable medium, wherein the non-transitory computer-readable medium is provisioned with program instructions that, when executed by at least one processor, cause a computing system to:

receive sound metadata that characterizes a plurality of sets of sound data;
generate a probabilistic model based on the received sound metadata, wherein the probabilistic model comprises a plurality of probability distribution functions for sound data characteristics that characterize the plurality of sets of sound data;
perform an acoustic simulation to obtain a noised version of an annotated speech sample, wherein performing the acoustic simulation comprises:
generating a scenario that comprises a set of sound data characteristics that are drawn from the probabilistic model, and
generating the noised version of the annotated speech sample that corresponds to the generated scenario by performing the acoustic simulation using the set of sound data characteristics;
and
simulate performance of a simulated network microphone device using the noised version of the annotated speech sample.

4 more elements

Dependencies are inferred from the claim text.

Systems and methods for generating labeled data to facilitate configuration of network microphone devices

Life of this patent

Systems and methods for generating labeled data to facilitate configuration of network microphone devices

Life of this patent

Abstract

Claims21