Abstract: An approach is described with respect to dynamically constructing and configuring a conversational agent learning model. Various aspects of the conversational agent learning model may be constructed and updated without continuous intervention of a domain administrator. A method pertaining to such approach may include retrieving a corpus of information. The corpus of information may include records from a set of repositories and external data, including data from social networks or applications. The method further may include configuring the conversational agent learning model based upon the retrieved corpus of information. The method further may include deploying the conversational agent learning model by facilitating interaction between the conversational agent learning model and a plurality of clients. The method further may include updating the conversational agent learning model to address any modification to the corpus of information.

Abstract: A method to detect a vocal command, the method including: analyzing audio data received from a transducer configured to convert audio into an electric signal and analyzing the data using a first neural network. The method also includes detecting a keyword from the audio data using the first neural network on the edge device, the first neural network being trained to recognize the keyword. The method further includes activating a second neural network after the keyword is identified by the first neural network and analyzing the audio data using the second neural network, the second neural network being trained to recognize a set of vocal commands. The method to detect a vocal command may also include detecting the vocal command word using the second neural network.

Abstract: A multimedia information processing method includes: parsing multimedia information to separate an audio from the multimedia information; converting the audio to obtain a mel spectrogram corresponding to the audio; determining, according to the mel spectrogram corresponding to the audio, an audio feature vector corresponding to the audio; and determining, based on an audio feature vector corresponding to a source audio in source multimedia information and an audio feature vector corresponding to a target audio in target multimedia information, a similarity between the target multimedia information and the source multimedia information.

Abstract: A method and device for fusing voiceprint features. The method includes: obtaining at least two voiceprint features of a voice sample of a target speaker (S3; S4); fusing the at least two voiceprint features on the basis of linear discriminant analysis (S5). The present method introduces a technique employing linear discriminant analysis to fuse various voiceprint features, so as to improve complementarities between the various voiceprint features and distinctions between the fused features, thereby increasing the recognition rate for target speakers and reducing the misrecognition rate for non-target speakers in voiceprint authentication scenarios, and providing personalized and improved user experience.

Abstract: Providing a predictive model for a target language by determining an instance weight for a labeled source language textual unit according to a set of unlabeled target language textual units, scaling, by the one or more computer processors, an error between a predicted label for the source language textual unit and a ground-truth label for the source language textual unit according to the instance weight, updating, by the one or more computer processors, network parameters of a predictive neural network model for the target language according to the error, and providing, by the one or more computer processors, the predictive neural network model for the target language to a user.

Abstract: Methods, systems, and computer program products for low-resource entity resolution with transfer learning are provided herein. A computer-implemented method includes processing input data via a first entity resolution model, wherein the input data comprise labeled input data and unlabeled input data; identifying one or more portions of the unlabeled input data to be used in training a neural network entity resolution model, wherein said identifying comprises applying one or more active learning algorithms to the first entity resolution model; training, using (i) the one or more portions of the unlabeled input data and (ii) one or more deep learning techniques, the neural network entity resolution model; and performing one or more entity resolution tasks by applying the trained neural network entity resolution model to one or more datasets.

Abstract: An encoder for providing an audio stream on the basis of a transform-domain representation of an input audio signal includes a quantization error calculator configured to determine a multi-band quantization error over a plurality of frequency bands of the input audio signal for which separate band gain information is available. The encoder also includes an audio stream provider for providing the audio stream such that the audio stream includes information describing an audio content of the frequency bands and information describing the multi-band quantization error. A decoder for providing a decoded representation of an audio signal on the basis of an encoded audio stream representing spectral components of frequency bands of the audio signal includes a noise filler for introducing noise into spectral components of a plurality of frequency bands to which separate frequency band gain information is associated on the basis of a common multi-band noise intensity value.

Abstract: The present invention provides an apparatus and method for automatic generation and update of a knowledge graph from multi-modal sources. The apparatus comprises a conversation parsing module configured for updating a dynamic information word set VD with labelled words generated from extracted from the multi-modal sources; updating a static information word set VS based on extracted schema of relations extracted from the multi-modal sources; and generating pairs of question and answer based on the dynamic information word set VD, the static information word set VS and the one or more sentence patterns; and a knowledge graph container configured for updating a knowledge graph based on the extracted entities of interest and schema of relations. Therefore, an efficient and cost-effective way for question decomposition, query chain construction and entity association from unstructured data is achieved.

Abstract: A processing system detects a period of non-voice activity and compares its duration to a cutoff period. The system adapts the cutoff period based on parsing previously-recognized speech to determine, according to a model, such as a machine-learned model, the probability that the speech recognized so far is a prefix to a longer complete utterance. The cutoff period is longer when a parse of previously recognized speech has a high probability of being a prefix of a longer utterance.

Abstract: Systems for analyzing and categorizing audio content that has been transcribed into text are provided. The systems include at least one machine that has a central processing unit, random access memory, a correlation module, a feature abstraction module, and at least one database. The correlation module is configured to receive written transcripts (each of which has been generated from audio content) and derive a correlation between each written transcript and one or more attributes. The feature abstraction module is configured to receive instructions that identify specific words within the written transcripts; replace the specific words with surrogate words; and associate correlative meanings with each of the surrogate words. The database is configured to receive, record, and make accessible to the feature abstraction module a table of specific words, each of which is associated with corresponding surrogate words and correlative meanings associated with each of the surrogate words.

Abstract: Methods and systems for converting an input content item into an output content item to enhance comprehension of the message by an interlocutor, based on contexts. For example, the conversion may occur in any message service: when an interlocutor writes a message in English (or any other language), he or she might include a regional dialect (purposively or not), such as a piece of slang, that the other interlocutors may not understand, although they all generally write and understand English. In such circumstances, the regional dialect is identified and replaced with either a more globalized word or with another linguistic regionalism that is understandable for the intended interlocutor.

Abstract: At an electronic device with a display, a microphone, and an input device: while the display is on, receiving user input via the input device, the user input meeting a predetermined condition; in accordance with receiving the user input meeting the predetermined condition, sampling audio input received via the microphone; determining whether the audio input comprises a spoken trigger; and in accordance with a determination that audio input comprises the spoken trigger, triggering a virtual assistant session.

Abstract: Techniques are described and relate to providing computing services using embeddings of a transformer-based encoder. In an example, a computer system generates, by using a machine learning (ML) transformer, an embedding vector based at least in part on text. The computer system stores the embedding vector and an association between the embedding vector and the text in a data store. Further, the computer system determines that a task is to be performed based at least in part on natural language understanding (NLU) of the text. The computer system receives the embedding vector from the data store based at least in part on the association between the embedding vector and the text. The task is performed based at least in part on the embedding vector after being received from the data store.

Abstract: A system described herein may provide a technique for the use of machine learning techniques to perform authentication, such as biometrics-based user authentication. For example, user biometric information (e.g., facial features, fingerprints, voice, etc.) of a user may be used to train a machine learning model, in addition to a noise vector. A representation of the biometric information (e.g., an image file including a picture of the user's face, an encoded file with vectors or other representation of the user's fingerprint, a sound file including the user's voice, etc.) may be iteratively transformed until the transformed biometric information matches the noise vector, and the machine learning model may be trained based on the set of transformations that ultimately yield the noise vector, when given the biometric information.

Abstract: A frame error concealment method based on frames including transform coefficient vectors including the following steps: It tracks sign changes between corresponding transform coefficients of predetermined sub-vectors of consecutive good stationary frames. It accumulates the number of sign changes in corresponding sub-vectors of a predetermined number of consecutive good stationary frames. It reconstructs an erroneous frame with the latest good stationary frame, but with reversed signs of transform coefficients in sub-vectors having an accumulated number of sign changes that exceeds a predetermined threshold.

Abstract: Provided is an artificial intelligence (AI) answering system including a user question receiver configured to receive a user question from a user terminal; a first question extender configured to generate a question template by analyzing the user question and determine whether the user question and the generated question template match; a second question extender configured to generate a similar question template by using a natural language processing and a deep learning model when the user question and the generated question template do not match; a training data builder configured to generate training data for training the second question extender by using an neural machine translation (NMT) engine; and a question answering unit configured to transmit a user question result derived through the first question extender or the second question extender to the user terminal.

Abstract: Implementations can receive audio data corresponding to a spoken utterance of a user, process the audio data to generate a plurality of speech hypotheses, determine an action to be performed by an automated assistant based on the speech hypotheses, and cause the computing device to render an indication of the action. In response to the computing device rendering the indication, implementations can receive additional audio data corresponding to an additional spoken utterance of the user, process the additional audio data to determine that a portion of the spoken utterance is similar to an additional portion of the additional spoken utterance, supplant the action with an alternate action, and cause the automated assistant to initiate performance of the alternate action. Some implementations can determine whether to render the indication of the action based on a confidence level associated with the action.

Abstract: Disclosed techniques relate to managing a dialogue between a user device and an autonomous agent. For instance, a computing device creates a discourse tree from a body of text that includes fragments. The discourse tree includes a root node, nonterminal nodes, and terminal nodes. Each nonterminal node represents a rhetorical relationship between two of the fragments and each terminal node is associated with one of the fragments. The computing device determines a main topic of the body of text from the discourse tree. The computing device provides the main topic to the user device, and in response, receives a user utterance. The computing device determines an intent from the user utterance and navigates the discourse tree to locate relevant information consistent with the intent.

Abstract: Techniques for generating natural language text with a natural language generation (NLG) system using a plurality of semantic objects including a first semantic object. The techniques include: obtaining a first specification of the first semantic object, the first specification specifying a first set of one or more data variables, first attributes, a first vocabulary, and a first document structure configuration; obtaining, from at least one data store, first data related to the first set of data variables; determining values of at least some of the first set of data variables using the first data; generating the natural language text including first natural language text, using the first specification, the values of at least some of the first set of data variables; and outputting the generated natural language text.

Abstract: A method for voiceprint recognition of an original speech is used to reduce information losses and system complexity of a model for data recognition of a speaker's original speech. The method includes: obtaining original speech data, and segmenting the original speech data based on a preset time length to obtain segmented speech data; performing tail-biting convolution processing and discrete Fourier transform on the segmented speech data through a preset convolution filter bank to obtain voiceprint feature data; pooling the voiceprint feature data through a preset deep neural network to obtain a target voiceprint feature; performing embedded vector transformation on the target voiceprint feature to obtain corresponding voiceprint feature vectors; and performing calculation on the voiceprint feature vectors through a preset loss function to obtain target voiceprint data, where the loss function includes a cosine similarity matrix loss function and a minimum mean square error matrix loss function.