Abstract: This disclosure describes techniques and systems for encoding instructions in audio data that, when output on a speaker of a first device in an environment, cause a second device to output content in the environment. In some instances, the audio data has a frequency that is inaudible to users in the environment. Thus, the first device is able to cause the second device to output the content without users in the environment hearing the instructions. In some instances, the first device also outputs content, and the content output by the second device is played at an offset relative to a position of the content output by the first device.

Abstract: A contextual end-to-end automatic speech recognition (ASR) system includes: an audio encoder configured to process input audio signal to produce as output encoded audio signal; a bias encoder configured to produce as output at least one bias entry corresponding to a word to bias for recognition by the ASR system; a transcription token probability prediction network configured to produce as output a probability of a selected transcription token, based at least in part on the output of the bias encoder and the output of the audio encoder; a first attention mechanism configured to receive the at least one bias entry and determine whether the at least one bias entry is suitable to be transcribed at a specific moment of an ongoing transcription; and a second attention mechanism configured to produce prefix penalties for restricting the first attention mechanism to only entries fitting a current transcription context.

Abstract: Techniques for computer system-based conversations are described. In an example, a system receives, from a first device, first data corresponding to a first interaction in a conversation that requests a function. The system causes the first device to output a first response to the first interaction. Prior to an execution of the function, the system determines that the conversation is to be paused and causes the first device to output a first indication that the conversation is paused. Upon determining that the conversation is to be resumed, the system causes a second device to output a second indication that the conversation is resumed. The second device can be the same or different from the first device. The system receives, from the second device, second data corresponding to a second interaction in the conversation and causes the execution of the function based at least in part on the second data.

Abstract: Devices and techniques are generally described for targeting of devices. In various examples, a first natural language input comprising a first request to output a response may be received by an input device. A first component may determine first data associated with the input device. A plurality of devices associated with the first data may be determined. First state data describing a state of each device of the plurality of devices may be determined. A first device of the plurality of devices may be determined as a target device for the first request based at least in part on the first state data. The first device may be different from the input device. First instructions may be sent to the first device effective to cause the first device to display the first visual content.

Abstract: A voice command resolution apparatus, including a memory configured to store instructions; and a processor configured to execute the instructions to: recognize a voice command of a user in an input sound, analyze a non-speech sound included in the input sound, and determine at least one target Internet of things (IoT) device related to execution of the voice command, based on an analysis result of the non-speech sound.

Abstract: In one aspect, a playback device includes a command-keyword engine having a local natural language unit (NLU). The playback device detects, via the command-keyword engine, a first command keyword in voice input of sound detected by one or more microphones of the playback device. The playback device determines whether the sound input data includes a keyword from a first predetermined library of keywords via a local natural language unit (NLU). The playback device transmits the input sound data to a second playback device over a local area network, the second playback device employing a second local NLU with a second predetermined library of keywords. The playback device receives a response from the second playback device and performs an action based on an intent determined by at least one of the first NLU or the second NLU according to the keywords in the voice input.

Abstract: An electronic apparatus includes a memory storing a speech recognition model and first recognition information corresponding to a first user voice obtained through the speech recognition model, the speech recognition model including a first network, a second network, and a third network; and a processor configured to: obtain a first vector by inputting voice data corresponding to a second user voice to the first network, obtain a second vector by inputting the first recognition information to the second network which generates a vector based on first weight information, and obtain second recognition information corresponding to the second user voice by inputting the first vector and the second vector to the third network which generates recognition information based on second weight information, wherein at least a part of the second weight information is the same as the first weight information.

Abstract: Embodiments described herein provide training a prompt generator for text classification. A first training dataset associated with a first plurality of class labels is received for a first training process. For a first instance of the first training dataset, a set of labels of interest is generated by sampling from a set of possible class labels including the first plurality of class labels. The prompt generator generates a first prompt based on the set of labels of interest. A pretrained language model generates a task output in response to an input of the first instance prepended with the first prompt. A loss objective is generated based on the task output and the set of labels of interest. Parameters of the prompt generator are updated based on the computed loss function via backpropagation while the pretrained language model is frozen.

Abstract: Techniques are described for training and/or utilizing sub-agent machine learning models to generate candidate dialog responses. In various implementations, a user-facing dialog agent (202, 302), or another component on its behalf, selects one of the candidate responses which is closest to user defined global priority objectives (318). Global priority objectives can include values (306) for a variety of dialog features such as emotion, confusion, objective-relatedness, personality, verbosity, etc. In various implementations, each machine learning model includes an encoder portion and a decoder portion. Each encoder portion and decoder portion can be a recurrent neural network (RNN) model, such as a RNN model that includes at least one memory layer, such as a long short-term memory (LSTM) layer.

Abstract: A system and method for controlling an electronic eyewear device using voice commands receives audio data from a microphone, processes the audio data to identify a wake word, and upon identification of a wake word, processes the audio data to identify at least one action keyword in the audio data. The audio data is provided to one of a plurality of controllers associated with different action keywords or sets of action keywords to implement an action. For example, the audio data may be provided to a settings controller to adjust settings of the electronic eyewear device when the action keyword is indicative of a request to adjust a setting of the electronic eyewear device or to a navigation controller to navigate to the system information of the electronic eyewear device when the action keyword is indicative of a request to navigate to system information of the electronic eyewear device.

Abstract: This application provides a speech signal processing method and apparatus, and relates to the field of signal processing technologies and earphone, to monitor an ambient sound signal and improve a monitoring effect and user experience. The method is applied to an earphone, where the earphone includes at least one external speech collector. The method includes: preprocessing a speech signal collected by the at least one external speech collector, to obtain an external speech signal; extracting an ambient sound signal from the external speech signal; and performing audio mixing processing on a first speech signal and the ambient sound signal based on amplitudes and phases of the first speech signal and the ambient sound signal and a location of the at least one external speech collector, to obtain a target speech signal.

Abstract: A system includes a development system and a digital assistance system. The development system includes a network interface configured to communicate with a plurality of communication channels, a processing system configured to interface with a project management subsystem, a scheduling subsystem, and the network interface, and an application programming interface configured to receive a command sequence for the project management subsystem and the scheduling subsystem. The digital assistance system includes a natural language processing engine configured to interface with a voice-enabled communication session through one of the communication channels. The digital assistance system also includes a command generator configured to generate the command sequence based on one or more requested tasks detected through the voice-enabled communication session and provide the command sequence to the application programming interface to execute the one or more requested tasks.

Abstract: A display device according to an embodiment of the present invention may include: a display unit which displays a content image; a microphone which receives a voice command of a user; a network interface unit for communicating with a natural language processing server and a search server; and a control unit which transmits the received voice command to the natural language processing server, receives intent analysis result information that indicates the intent of the user, which corresponds to the voice command, from the natural language processing server, and performs a function of the display device according to the received intent analysis result information.

Abstract: A learning device includes a memory, and processing circuitry coupled to the memory and configured to receive an input of a plurality of series for learning having known accuracy, and learn a model represented by a neural network, the model being capable of determining accuracy levels of two series when given feature amounts of the two series among the plurality of series.

Abstract: System and method of generating an executable action item in response to natural language dialogue are disclosed herein. A computing system receives a dialogue message from a remote client device of a customer associated with an organization, the dialogue message comprising an utterance indicative of an implied goal. A natural language processor of the computing system parses the dialogue message to identify one or more components contained in the utterance. The planning module of the computing system identifies the implied goal. The computing system generates a plan within a defined solution space. The computing system generates a verification message to the user to confirm the plan. The computing system transmits the verification message to the remote client device of the customer. The computing system updates an event queue with instructions to execute the action item according to the generated plan upon receiving a confirmation message from the remote client device.

Abstract: A method, computer program product, and computing system for defining a communication computing system within a computing network, wherein the computing network includes a plurality of disparate platforms configured to provide information concerning various topics; enabling a user to issue a verbal command concerning one or more of the plurality of disparate platforms; processing the verbal command to generate a platform-useable command based, at least in part, upon the verbal command; and providing the platform-useable command to at least a portion of the plurality of disparate platforms via the communication computing system.

Abstract: Embodiments provide a voice interaction method and an apparatus, and relate to the field of terminal technologies. Common voice skill commands in a first application scenario may be determined based on the first application scenario and a historical voice skill usage record, and displayed in a display interface. This can implement scenario-based recommendation of voice skill commands, to cover as many application scenarios as possible. In this application, after being woken up, the voice assistant determines the first application scenario based on one or more information items. The voice assistant determines the common voice skill commands in the first application scenario based on the first application scenario and the historical voice skill usage record. The voice assistant displays the common voice skill commands in the first application scenario in the display interface.

Abstract: In one example, a method being performed by a computer system comprises: receiving an image file containing a pathology report; performing an image recognition operation on the image file to extract input text strings; detecting, using a natural language processing (NLP) model, entities from the input text strings, each entity including a label and a value; extracting, using the NLP model, the values of the entities from the input text strings; converting, based on a mapping table that maps entities and values to pre-determined terminologies, the values of at least some of the entities to the corresponding pre-determined terminologies; and generating a post-processed pathology report including the entities detected from the input text strings and the corresponding pre-determined terminologies.

Abstract: A computer device acquires speech content. The device performs feature extraction on the speech content to obtain an intermediate feature. The intermediate feature is used for indicating an audio expression characteristic of the speech content. The device decodes the intermediate feature based on an attention mechanism to obtain a first word graph network. The device performs feature mapping on the intermediate feature based on pronunciation of the speech content to obtain a second word graph network. The device determines a recognition result of the speech content according to candidate word connection relationships indicated by the first word graph network and the second word graph network.

Abstract: A system and method for machine learning classification of user sentiment is disclosed. The method includes storing including a plurality of category information. The plurality of category information includes a set of domain-specific category information. The method further includes extracting a plurality of aspects from textual data. The method further includes generating a sentiment by a machine learning model. The method further includes receiving the plurality of aspects and the set of domain-specific category information. The method further includes generating a sentiment based on the plurality of aspects and the set of domain-specific category information.