Abstract: The example embodiments are directed toward improvements in document classification. In an embodiment, a method is disclosed comprising generating a set of sentences based on a document; predicting a set of labels for each sentence using a multi-label classifier, the multi-label classifier including a self-attended contextual word embedding backbone layer, a bank of trainable unigram convolutions, a bank of trainable bigram convolutions, and a fully connected layer the multi-label classifier trained using a weakly labeled data set; and labeling the document based on the set of labels. The various embodiments can target multiple use cases such as identifying related entities, trending related entities, creating ephemeral timeline of entities, and others using a single solution. Further, the various embodiments provide a weakly supervised framework to train a model when a labeled golden set does not contain a sufficient number of examples.

Abstract: A method including transcribing, automatically, an ongoing stream of voice data into text phrases. The method also includes receiving an indication of a selected text phrase in the text phrases. The method also includes converting the selected text phrase to a selected phrase vector. The method also includes generating a subsequent text phrase, after the selected text phrase, from the ongoing stream of voice data, and adding the subsequent text phrase to the text phrases. The method also includes converting the subsequent text phrase to a subsequent phrase vector. The method also includes generating a similarity confidence score from the selected phrase vector and the subsequent phrase vector, using a machine learning model. The method also includes highlighting, responsive to the similarity confidence score exceeding a threshold value, the subsequent text phrase in the text phrases.

Abstract: A system for receiving a corpus of sign language data in which a plurality of known signs each correspond to known meanings, generate a model for identifying new sign language signs using the corpus, and identifying, using the model, a new sign language sign that does not match any of the plurality of known signs.

Abstract: Implementations relate to proactively aggregating client device data to append to client assistant data that is communicated to a server device in response to a user request to a client automated assistant. When a user request that is associated with, for example, a peripheral client device, is received at a client device, the client device can communicate, to a server device, data that embodies the user request (e.g., audio data and/or local speech recognition data), along with peripheral device data that was received before the client device received the user request. In this way, the client automated assistant can bypass expressly soliciting peripheral device data each time a user request is received at another client device. Instead, a peripheral device can proactively communicate device data to a client device so that the device data can be appended to request data communicated to the server device from a particular client device.

Abstract: The present invention relates to a method for voice prompting, a mobile control terminal and a system for voice prompting. The method includes: receiving, by the mobile control terminal, a warning instruction transmitted by an unmanned aerial vehicle (UAV); generating corresponding text information according to the warning instruction; converting the text information to an audio file; and playing the audio file. The mobile control terminal of the present invention provides various warning information to a user by playing an audio file, which is not a preset audio file directly stored in the mobile control terminal, but an audio file converted from text information generated by the mobile control terminal according to a warning instruction transmitted by the UAV. Therefore, there is no need to store a preset audio file in the mobile control terminal, which reduces a memory space occupied in the mobile control terminal and improves performance of the mobile control terminal.

Abstract: A method for computer control of online social interactions based on conversation processing. A portion of a conversation with a given user is recorded, stored in memory, and processed so as to influence subsequent interactions. These may include audio or other contextualized annunciations of subsequent telephone calls. Other modes addressing conversational processing for social purposes can result in interwoven conversation guided among threads of interest, advertising and incented participation of conversational content and placement. The invention is capable of implementation in telecommunications systems such as cellular, local exchange, and VOIP, and in combination with other forms of internet-based telecommunication, including smart phones and adaptive forums chat rooms.

Abstract: A data labeling method and device and a computer-readable storage medium. The method includes: based on a knowledge graph, remotely acquiring data to be labeled; performing data cleaning and pre-labeling to the data to be labeled, to obtain pre-labeled data; and performing labeling correction to the pre-labeled data.

Abstract: A method according to one embodiment includes receiving audio input by a microphone of an access control device that controls access through a passageway, processing an audio signal associated with the audio input to identify and authenticate a user, determining a command corresponding with the audio signal in response to identification and authentication of the user, and performing at least one action that corresponds with the command.

Abstract: A Unified Speech and Audio Codec (USAC) that may process a window sequence based on mode switching is provided. The USAC may perform encoding or decoding by overlapping between frames based on a folding point when mode switching occurs. The USAC may process different window sequences for each situation to perform encoding or decoding, and thereby may improve a coding efficiency.

Abstract: A voice interaction method, where a service type set on which a user has a voice interaction intention is predicted based on a target event that can trigger voice interaction, and when a service type of a first service expressed by a voice instruction is a target service type in the service type set, the first service is executed.

Abstract: A system and method for contextualized communication events and notifications for project management based on conversation processing. A portion of a conversation with a given user is recorded, stored in memory, and processed so as to influence subsequent interactions. These may include audio or other contextualized annunciations of related information and expectations concerning project responsibilities, schedules, and dependencies. Subsequent meetings and telephone calls may chivvy the project along through reminders of potential compliance or noncompliance with stated and accepted assignments. Other modes addressing conversational processing for social purposes can result in interwoven conversation guided among threads of interest, advertising and incented participation of conversational content and placement.

Abstract: Disclosed is a method for a social interaction by a robot device. The method includes receiving an input from a user, determining an emotional state of the user by mapping the received input with a set of emotions and dynamically interacting with the user based on the determined emotional state in response to the input. Dynamically interacting with the user includes generating contextual parameters based on the determined emotional state. The method includes determining an action in response to the received input based on the generated contextual parameters and performing the determined action. The method further includes receiving another input from the user in response to the performed action and dynamically updating the mapping of the received input with the set of emotions based on the other input for interacting with the user.

Abstract: An apparatus for selecting one of a first encoding algorithm having a first characteristic and a second encoding algorithm having a second characteristic for encoding a portion of an audio signal to obtain an encoded version of the portion of the audio signal has a first estimator for estimating a first quality measure for the portion of the audio signal, which is associated with the first encoding algorithm, without actually encoding and decoding the portion of the audio signal using the first encoding algorithm. A second estimator is provided for estimating a second quality measure for the portion of the audio signal, which is associated with the second encoding algorithm, without actually encoding and decoding the portion of the audio signal using the second encoding algorithm. The apparatus has a controller for selecting the first or second encoding algorithms based on a comparison between the first and second quality measures.

Abstract: An artificial intelligence enabled system is disclosed. The system includes a core component for enabling AI-powered interactions between the system and its users and one or more agents that understand user intent and automatically interact with products and services on the web and/or in the physical world through imitation of a human user.

Abstract: A semantic parsing method includes extracting, by a first server, a first entity of a first slot from a first user sentence, modifying the first entity to a second entity of a second slot when the first entity is a pronoun, sending a first server request to a second server, and obtaining a first service result from the second server in response to the first service request. The first service request includes indication information of a first intention and the first entity, the second server is an application server that provides a first skill, and the first service result is based on the first intention and the first entity. The first server returns the first service result to a device to output by the device.

Abstract: In accordance with an example embodiment of the present invention, disclosed is a method and an apparatus for voice activity detection (VAD). The VAD comprises creating a signal indicative of a primary VAD decision and determining hangover addition. The determination on hangover addition is made in dependence of a short term activity measure and/or a long term activity measure. A signal indicative of a final VAD decision is then created.

Abstract: Methods, apparatus, system and computer-implemented method are provided for embedding a portion of text describing one or more entities of interest and a relationship. The portion of text describes a relationship for the one or more entity(ies) of interest, where the portion of text includes multiple separable entities describing the relationship and the entity(ies). The multiple separable entities including the one or more entity(ies) of interest and one or more relationship entity(ies). A set of embeddings for each of the separable entities is generated, where the set of embeddings for a separable entity includes an embedding for the separable entity and an embedding for at least one entity associated with the separable entity. One or more composite embeddings may be formed based on at least one embedding from each of the sets of embeddings. The composite embedding(s) may be sent for input to a machine learning model or classifier.

Abstract: An apparatus for estimating an inter-channel time difference between a first channel signal and a second channel signal, includes: a calculator for calculating a cross-correlation spectrum for a time block from the first channel signal in the time block and the second channel signal in the time block; a spectral characteristic estimator for estimating a characteristic of a spectrum of the first channel signal or the second channel signal for the time block; a smoothing filter for smoothing the cross-correlation spectrum over time using the spectral characteristic to obtain a smoothed cross-correlation spectrum; and a processor for processing the smoothed cross-correlation spectrum to obtain the inter-channel time difference.

Abstract: Systems and methods are disclosed for tuning a generative artificial intelligence (AI) model based on a knowledge base. Instead of manually generating questions relevant to the knowledge base, providing those questions to the generative AI model, and manually reviewing the answers generated by the generative AI model in order to tune the generative AI model over many iterations, a natural language processing model may be configured to leverage the knowledge base to automatically generate questions and answers based on the knowledge base. In this manner, the natural language processing model is able to generate tuning data that may be used to automatically tune the generative AI model. The systems and methods also disclose automatic tuning of the generative AI model, including testing and feedback that may be used to improve tuning of the generative AI model.

Abstract: An information processing apparatus includes an extracting unit (133) that extracts a changing message related to a change in macro data (M), the changing message including at least one piece of first information indicating a function to be executed, and second information linked to the first information, from a user speech; a presuming unit (134) that presumes an element to be changed in the macro data (M) based on the changing message extracted by the extracting unit (133); and a changing unit (135) that changes the element to be changed in the macro data (M) presumed by the presuming unit (134), based on the changing message.