Methods and apparatus for reducing latency in speech recognition applications
AssigneeNuance Communications, Inc.
InventorMark Fanty
Life of this patent
Estimating the term…
Abstract
The method comprises receive first audio comprising speech from a user of a computing device, detecting an end of speech in the first audio, generating an ASR result based, at least in part, on a portion of the first audio prior to the detected end of speech, determining whether a valid action can be performed by a speech-enabled application installed on the computing device using the ASR result, and processing second audio when it is determined that a valid action cannot be performed by the speech-enabled application using the ASR result.
Claims19
An apparatus comprising:
- at least one processor programmed to:
- determine whether an automatic speech recognition (ASR) result, generated by an ASR engine based at least in part on received audio of speech of a user, includes any of one or more prefixes for the user, each of the prefixes being associated with a corresponding threshold time usable by the ASR engine for endpointing, and,
- if it is determined that the ASR result includes at least one prefix of the one or more prefixes stored in a memory, for each of the at least one prefix in the ASR result, update an endpointing wait time used by the ASR engine for endpointing,
- wherein the at least one processor causes the ASR engine to perform dynamic endpointing by instructing the ASR engine to change the endpointing wait time for each of the at least one prefix in the ASR result to the corresponding threshold time associated with the one or more prefixes, respectively.
A method of a speech processing device, the method comprising:
- using at least one processor to:
- determine whether an automatic speech recognition (ASR) result, generated by an ASR engine based at least in part on received audio of speech of a user, includes any of one or more prefixes for the user, each of the one or more prefixes being associated with a corresponding threshold time useable by the ASR engine for endpointing,
- update an endpointing wait time used by the ASR engine for endpointing for each of the one or more prefixes in the ASR result, if it is determined that the ASR result includes at least one prefix of the one or more prefixes, and
- for each prefix of the at least one prefix in the ASR result, respectively, perform dynamic endpointing by instructing the ASR engine to change the endpointing wait time to the corresponding threshold time associated with the prefix.
At least one non-transitory computer-readable storage medium encoded with a plurality of instructions that, when executed by at least one processor of a speech processing device, cause the at least one processor to perform a method comprising:
- determining whether an automatic speech recognition (ASR) result, generated by an ASR engine based at least in part on received audio of speech of a user, includes any of one or more prefixes for the user, the one or more prefixes being associated with a corresponding threshold time useable by the ASR engine for endpointing,
- updating an endpointing wait time used by the ASR engine for endpointing for each of the one or more prefixes in the ASR result, if it is determined that the ASR result includes at least one prefix of the one or more prefixes, and
- for each prefix of the at least one prefix in the ASR result, respectively, performing dynamic endpointing by instructing the ASR engine to change the endpointing wait time to the corresponding threshold time associated with the prefix.
Dependencies are inferred from the claim text.