Amazon Technologies, Inc.

Google Inc.

Yuzo Watanabe

Matthew Paul Bell

Stephen Michael Polansky

Arushan Rajasekaram

Isaac Scott Noble

Nikko Strom

Madan Mohan Rao Jampani

Hartwig Adam

Li Zhang

Kenneth Mark Karakotsios

Rohit Prasad

Ariya Rastrow

William Folwell Barton

Stan Weidner Salvador

Stephen Frederick Potter

David Karam

Gabrielle M Halberg

Ariel Gilder

Eric Penner

Farooq Ahmad

Bjorn Hoffmeister

Mark Steven Baumback

Rajiv Ramachandran

Ryan H Cassidy

Paul Joseph Schaffert

The present disclosure is directed to processing imagery using one or more machine learning (ML) models. In particular, data describing imagery comprising a plurality of different and distinct frames can be received; and based at least in part on one or more ML models and the data describing the imagery, and for each frame of the plurality of different and distinct frames, one or more scores can be determined for the frame. Each score of the score(s) can indicate a determined measure of suitability of the frame with respect to one or more of various different and distinct uses for which the ML model(s) are configured to determine suitability of imagery.

Methods and systems for processing imagery

Described herein are systems, methods, and apparatus for determining audio context between an audio source and an audio sink and selecting signal profiles based at least in part on that audio context. The signal profiles may include noise cancellation which is configured to facilitate operation within the audio context. Audio context may include user-to-user and user-to-device communications.

Signal processing based on audio context

Features are disclosed for generating predictive personal natural language processing models based on user-specific profile information. The predictive personal models can provide broader coverage of the various terms, named entities, and/or intents of an utterance by the user than a personal model, while providing better accuracy than a general model. Profile information may be obtained from various data sources. Predictions regarding the content or subject of future user utterances may be made from the profile information. Predictive personal models may be generated based on the predictions. Future user utterances may be processed using the predictive personal models.

Generation of predictive natural language processing models

A speech-based audio device may be configured to detect a user-uttered trigger expression and to respond by interpreting subsequent words or phrases as commands. In order to distinguish between utterance of the trigger expression by the user and generation of the trigger expression by the device itself, output signals used as speaker inputs are analyzed to detect whether the trigger expression has been generated by the speaker. If a detected trigger expression has been generated by the speaker, it is disqualified. Disqualified trigger expressions are not acted upon the by the audio device.

Qualifying trigger expressions in speech-based systems

Described herein are systems and methods for the generation and maintenance of language models. Language models are developed based at least in part on transaction data from one or more users. These transactions may include purchases and other interactions between one or more users and one or more online merchants. The language models may be associated with a particular user or group of users. Ongoing transactions may modify the language models. The language models may be used to recognize spoken input from the one or more users.

Generation and maintenance of language model

Described herein are systems and methods for context-aware speech processing. A speech context is determined based on context data associated with a user uttering speech. The speech context and the speech uttered in that speech context may be used to build acoustic models for that speech context. An acoustic model for use in speech processing may be selected based on the determined speech context. A language model for use in speech processing may also be selected based on the determined speech context. Using the acoustic and language models, the speech may be processed to recognize the speech from the user.

Context-aware speech processing

A system for controlling multiple devices using automatic speech recognition (ASR) even when the devices may not be capable of performing ASR themselves. A device such as a media player, appliance, or the like may be recognized by a network. The configured controls for the device (such as a remote control or other mechanism) are incorporated into a device control registry which catalogs device command controls. Individual ASR grammars are constructed for the devices so speech commands for those devices may be processed by an ASR device. The ASR device may then process those speech commands and convert them into the appropriate inputs for the controlled device. The inputs may then be sent to the controlled device, resulting in ASR control for non-ASR devices.

Speech recognition capability generation and control

Various approaches enable automatic communication generation based on patterned behavior in a particular context. For example, a computing device can monitor behavior of a user to determine patterns of communication behavior in certain situations. In response to detecting multiple occurrences of the certain situation, a computing device can prompt a user to perform an action corresponding to the pattern of behavior. In some embodiments, a set of speech models corresponding to a type of contact is generated. The speech models include language consistent with patterns of speech between a user and the type of contact. Based on context and on the contact, a message using language consistent with past communications between the user and contact is generated from a speech model associated with the type of contact.

Customized speech generation

Features are disclosed for maintaining data that can be used to personalize spoken language processing, such as automatic speech recognition ('ASR'), natural language understanding ('NLU'), natural language processing (&#x201c;NLP&#x201d;), etc. The data may be obtained from various data sources, such as applications or services used by the user. User-specific data maintained by the data sources can be retrieved and stored for use in generating personal models. Updates to data at the data sources may be reflected by separate data sets in the personalization data, such that other processes can obtain the update data sets separate from other data.

Retrieval and management of spoken language understanding personalization data

A computing device can capture audio data representative of audio content present in a current environment. The captured audio data can be compared with audio models to locate a matching audio model. The matching audio model can be associated with an environment. The current environment can be identified based on the environment associated with the matching audio model. In some embodiments, information about the identified current environment can be provided to at least one application executing on the computing device. The at least one the application can be configured to adjust at least one functional aspect based at least in part upon the determined current environment. In some embodiments, one or more computing tasks performed by the computing device can be improved based on information relating to the identified current environment. These computing tasks can include location refinement, location classification, and speech recognition.

Growing community of inventors

Seattle, WA, United States of America

Yuzo Watanabe