US 10,832,668 B1 · Dynamic speech processing

Abstract

Techniques for dynamically maintaining speech processing data on a local device for frequently input commands are described. One or more devices receive speech processing data specific to one or more commands associated with system input frequencies satisfying an input frequency threshold. The device(s) then receives input audio corresponding to an utterance and generate input audio data corresponding thereto. The device(s) performs speech recognition processing on input audio data to generate input text data using a portion of the received speech processing data. The device(s) determines a probability score associated with the input text data and determines the probability score satisfies a threshold probability score. The device(s) then performs natural language processing on the input text data to determine the command using a portion of the speech processing data. The device(s) then outputs audio data responsive to the command.

Claims20

Dependent claims

A computer-implemented method performed by a user device, comprising:

receiving speech processing model data from a remote device, the speech processing model data being received based at least in part on usage history data representing a first speech input was received at least a number of times over a duration of time;
after receiving the speech processing model data, receiving a second speech input;
using the speech processing model data, performing speech recognition processing on first audio data, corresponding to the second speech input, to generate first speech recognition output data;
determining a first confidence score representing the first speech recognition output data corresponds to the first speech input;
determining the first confidence score fails to satisfy a threshold confidence score;
after determining the first confidence score fails to satisfy the threshold confidence score, sending the first audio data to the remote device;
receiving, from the remote device, first output data responsive to the second speech input;
and
causing a first output corresponding to the first output data.

6 more elements

A user device comprising:

at least one processor;
and
at least one memory including instructions that, when executed by the at least one processor, cause the user device to:
receive speech processing model data from a remote device, the speech processing model data being received based at least in part on usage history data representing a first speech input was received at least a number of times over a duration of time, the speech processing model data comprising at least a first portion for use in speech recognition processing;
after receiving the speech processing model data, receive a second speech input;
using the speech processing model data, perform speech recognition processing on first audio data corresponding to the second speech input to generate first speech recognition output data;
determine a first confidence score representing the first speech recognition output data corresponds to the first speech input;
determine the first confidence score fails to satisfy a threshold confidence score;
send the first speech recognition output data to the remote device;
receive, from the remote device, first output data responsive to the second speech input;
and
cause a first output corresponding to the first output data.

9 more elements

A computer-implemented method performed by a user device, comprising:

receiving speech processing model data from a remote device, the speech processing model data being received based at least in part on usage history data representing a first speech input was received at least a number of times over a duration of time, the speech processing model data comprising at least a first portion for use in speech recognition processing;
after receiving the speech processing model data, receiving a second speech input;
using the speech processing model data, performing speech recognition processing on first audio data, corresponding to the second speech input, to generate first speech recognition output data;
determining a first confidence score representing the first speech recognition output data corresponds to the first speech input;
determining the first confidence score fails to satisfy a threshold confidence score;
generating an indicator representing the speech recognition processing of the first audio data was unsuccessful;
sending the indicator to the remote device;
and
after sending the indicator, receiving second speech processing model data from the remote device.

6 more elements

Dependencies are inferred from the claim text.

Dynamic speech processing

Life of this patent

Dynamic speech processing

Life of this patent

Abstract

Claims20