Abstract: Implementations relate to selecting a particular device, from an ecosystem of devices, to provide responses to a device-agnostic request of the user while a scenario is occurring. The user specifies a scenario and contextual features are identified from one or more devices of the ecosystem to generate scenario features indicative of the scenario occurring. The scenario features are stored with a correlation to a device that is specified by the user to handle responses while the scenario is occurring. When a subsequent device-agnostic request is received, current contextual features are identified and compared to the scenario features. Based on the comparison, the specified assistant device is selected to respond to the device-agnostic request.

Abstract: In some implementations, a device may obtain a natural language input including an emoji. The device may identify one or more appearance modifiers associated with the emoji. The device may generate a token associated with the emoji that removes the one or more appearance modifiers, wherein the token is associated with multiple emojis including the emoji, and wherein the token is a modified code associated with the emoji or is associated with a cluster that is associated with the multiple emojis. The device may provide, to a natural language processing (NLP) model, the token associated with the emoji. The device may obtain, from the NLP model, an output that indicates an interpretation of the natural language input based on providing the token to the NLP model.

Abstract: Approaches presented herein can provide for the performance of specific types of tasks using a large model, without a need to retrain the model. Custom endpoints can be trained for specific types of tasks, as may be indicated by the specification of one or more guidance mechanisms. A guidance mechanism can be added to or used along with a request to guide the model in performing a type of task with respect to a string of text. An endpoint receiving such a request can perform any marshalling needed to get the request in a format required by the model, and can add the guidance mechanisms to the request by, for example, prepending one or more text strings (or text prefixes) to a text-formatted request. A model receiving this string can process the text according to the guidance mechanisms. Such an approach can allow for a variety of tasks to be performed by a single model.

Abstract: A method of generating summaries of content items using one or more large language models (LLMs) is disclosed. A first content item is identified. The first content item includes a set of sub-content items. A level of abstraction is determined for the content item. A prompt is automatically engineered for providing to the one or more LLMs. The prompt includes a reference to the first content item and the level of the abstraction for the first content item. A response to the prompt is received from the LLM. The response includes a second content item. The second content item includes a representation of the first content item that is generated by the LLM. The representation omits or simplifies one or more of the set of sub-content items based on the level of abstraction. The representation is used to control an output that is communicated to a target device.

Abstract: An electronic device includes a memory storing first vector information obtained from a pre-registered user voice, and a processor configured to obtain, based on a user voice being received, second vector information of a first filtered user voice by inputting the received user voice and the first vector information stored in the memory to a trained first neural network model, obtain second filtered user voice information by inputting the second vector information of the first filtered user voice and the received user voice to a trained second neural network model, and perform voice recognition based on the second filtered user voice information.

Abstract: A method includes receiving a training example that includes audio data representing a spoken utterance and a ground truth transcription. For each word in the spoken utterance, the method also includes inserting a placeholder symbol before the respective word identifying a respective ground truth alignment for a beginning and an end of the respective word, determining a beginning word piece and an ending word piece, and generating a first constrained alignment for the beginning word piece and a second constrained alignment for the ending word piece. The first constrained alignment is aligned with the ground truth alignment for the beginning of the respective word and the second constrained alignment is aligned with the ground truth alignment for the ending of the respective word. The method also includes constraining an attention head of a second pass decoder by applying the first and second constrained alignments.

Abstract: As described herein, various embodiments of the present invention provide methods, apparatus, systems, computing devices, computing entities, and/or the like for performing natural language processing operations for generating guided summaries using summarization templates that are mapped to hybrid classes of a hybrid classification space for a hybrid classification machine learning model. In some embodiments, by using summarization templates, a proposed summarization framework is able to vastly reduce the computational complexity of performing summarization on an input document data object, such as an input multi-party communication transcript data object, by defining the set of dynamic data fields that apply to the input document data object based at least in part on an assigned class/category of the input document data object.

Abstract: Provided is a technique according to which it is possible to obtain a decoded sound signal of high sound quality without significantly increasing the delay time compared to a configuration in which only a decoded sound signal of the minimum necessary sound quality is obtained. In a terminal apparatus connected to a first communication line and a second communication line with a lower priority level there than, sound signals of multiple channels are obtained and output based on a monaural code included in a first code string input from the first communication line and an extended code included in a second code string with the closest frame number to that of the monaural code among extended codes included in the second code string input from the second communication line.

Abstract: One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to a process to facilitate multi-lingual query interpretation. A system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an annotation component that generates one or more language invariant signals, an interpretation component that generates a complete query intent using the one or more language invariant signals, and a translation component that processes the complete query intent to an executable backend query to facilitate multi-lingual query interpretation. In one or more embodiments, the translation component can be operatively connected with the interpretation component to generate a zero-shot transfer of the one or more language invariant signals.

Abstract: Modern automatic speech recognition (ASR) systems can utilize artificial intelligence (AI) models to service ASR requests. The number and scale of AI models used in a modern ASR system can be substantial. The process of configuring and reconfiguring hardware to execute various AI models corresponding to a substantial number of ASR requests can be time consuming and inefficient. Among other features, the described technology utilizes batching of ASR requests, splitting of the ASR requests, and/or parallel processing to efficiently use hardware tasked with executing AI models corresponding to ASR requests. In one embodiment, the compute graphs of ASR tasks are used to batch the ASR requests. The corresponding AI models of each batch can be loaded into hardware, and batches can be processed in parallel. In some embodiments, the ASR requests are split, batched, and processed in parallel.

Abstract: Artificial intelligence (AI) technology can be used in combination with composable communication goal statements to facilitate a user's ability to quickly structure story outlines in a manner usable by an NLG narrative generation system without any need for the user to directly author computer code. Narrative analytics that are linked to communication goal statements can employ a conditional outcome framework that allows the content and structure of resulting narratives to intelligently adapt as a function of the nature of the data under consideration. This AI technology permits NLG systems to determine the appropriate content for inclusion in a narrative story about a data set in a manner that will satisfy a desired communication goal.

Abstract: A system for speech interpretation from a users' speech, while in a virtual environment, aided by user data and virtual world data. This system includes a virtual reality device comprising one or more user input devices, one or more user output devices, and a communication module. The output devices outputting a virtual environment to the user. A database stores information about elements in the virtual environment. An artificial intelligence module performs speech interpretation. The artificial intelligence module comprises a speech-to-text module that interprets user speech into a plurality of textual interpretations, and based on a ranking of the textual interpretations, select a top interpretation. An augmentation module adds context into the user speech to aid interpreting the speech. The context is derived from user data regarding the user's interaction with the virtual environment, and virtual environment data defining an element in the virtual environment with which the user is interacting.

Abstract: Filtering user intents corresponding to user utterances is provided. A list of allowed user intents is generated, using a natural language understanding model of a chatbot, based on identifying one or more of a set of user intents corresponding to a user utterance within a filtered user intent mapping table. It is determined whether a user intent having a highest confidence score in the set of user intents corresponding to the user utterance is contained in the list of allowed user intents. In response to determining that the user intent having the highest confidence score in the set of user intents corresponding to the user utterance is contained in the list of allowed user intents, content corresponding to the user intent having the highest confidence score is sent, using the chatbot, to a client device of a user who submitted the user utterance as a response to the user utterance.

Abstract: An information processing apparatus includes an acquisition unit that acquires, from a storage unit that stores episode data of a speaker, the episode data regarding topic information included in utterance data of the speaker. The information processing apparatus further includes an interaction control unit that controls an interaction with the speaker so as to include an episode based on the episode data.

Abstract: A target signal extraction apparatus according to an embodiment of the present invention may comprise a steering vector estimator and a beamformer. The steering vector estimator may generate an input signal covariance according to input results for each frequency over time, generate a noise covariance on the basis of a variance determined according to output results corresponding to the input results, and generate a steering vector on the basis of the input signal covariance and the noise covariance. The beamformer may generate a beamforming weight according to a beamforming covariance determined according to the variance and the steering vector, and provide the output results on the basis of the input results and the beamforming weight.

Abstract: A schema and cell value aware Named Entity Recognition (NER) model is used to perform natural language queries. Natural language queries may be received via an interface of a natural language query processing system. A fuzzy search may be performed that allows non-exact matches for column names or cell values of data sets potentially used to answer the natural language query. An NER model that adds a type embedding for an exact match of a column name or cell found in the fuzzy search that corresponds to a span of one or more words may be applied as part of generating the entity prediction for the natural language query. One or more queries to at least one of the data sets may be performed to return a result to the natural language query using the entity prediction generated by the NER machine learning model.

Abstract: Systems and methods are provided that include a processor executing a program to receive input text, divide the input text into sentences, identify one or a plurality of jurisdiction candidates in the sentences from a predetermined taxonomy to generate a jurisdictions list, transform the jurisdictions list using a type recognition neural network to disambiguate jurisdictions in the jurisdictions list, the type recognition neural network being trained on a labeled ground truth dataset containing pairs of geographic names and tax jurisdiction types, and generate and output the jurisdictions list as a jurisdiction prediction list.

Abstract: A method and system for automating a process of downloading and analyzing messages from conversation rooms and chat rooms to determine topics, entities, context, and actionable items are provided. The method includes downloading a set of messages that have been communicated over a communication channel; analyzing each respective message in order to determine at least one respective topic that relates to each respective message; determining, based on a result of the analysis, metrics that relate to the set of messages; and storing historical data that relates to the downloaded set of messages and each of the metrics. The analysis may be performed by executing an artificial intelligence (AI) algorithm that is based on a Natural Language Processing (NLP) model and is trained by using the historical data.

Abstract: An information processing apparatus includes an acquisition unit that acquires, from a storage unit that stores episode data of a speaker, the episode data regarding topic information included in utterance data of the speaker. The information processing apparatus further includes an interaction control unit that controls an interaction with the speaker so as to include an episode based on the episode data.

Abstract: Techniques for providing natural language understanding (NLU) services to contact centers are disclosed herein. An example method includes receiving, at a cloud NLU connector, a data stream from a contact center that includes an audio stream, extracting the audio stream from the data stream, and transmitting the audio stream through a secure network to a cloud-based NLU hub. The cloud-based NLU hub is communicatively coupled to a plurality of cloud-based NLU service providers and stores a plurality of user NLU profiles that each designate one or more cloud-based NLU service providers to provide NLU services. The example method further includes determining one or more designated cloud-based NLU service providers to provide at least one NLU service corresponding to the audio stream, and causing the one or more designated cloud-based NLU service providers to process the audio stream in accordance with the at least one NLU service.