Abstract: Systems and methods that access an online networked resource using a locator are disclosed. A first item of content on the networked resource is identified. A trigger rule comprising keywords and a sentiment classifier is accessed. A neural network, including input, hidden, and output layers, is used to assign a sentiment classification to the first item of content. The trigger rule, the sentiment classification, and identified keywords, are used to determine whether response content is to be posted to the online networked resource. In response to determining, using the trigger rule, the assigned sentiment classification, and keywords identified in the first item of content, that response content is to be posted to the online networked resource, the sentiment classification and identified keywords are used to select and/or generate a second item of content, and the second item of content is enabled to be posted to the online networked resource.

Abstract: Provided are methods and systems to verify user identity for voice enabled devices. A voice input can instruct a voice enabled device to perform a plurality of functions/services that, depending on the function/service, may require additional user verification. Primary user verification can be performed by associating voice characteristics of the voice input to a profile associated with a user/user device. A signal (e.g., a BLE beacon) can be sent to the user device that causes the user device to perform secondary user verification. The secondary user verification can be based on a biometric input, passcode verification, authenticated message reply, for example. Based on the secondary user verification, an operational command associated with the voice input can be executed.

Abstract: The invention provides an efficient implementation of cross-product enhanced high-frequency reconstruction (HFR), wherein a new component at frequency Q?+r?0 is generated on the basis of existing components at ? and ?+?0. The invention provides a block-based harmonic transposition, wherein a time block of complex subband samples is processed with a common phase modification. Superposition of several modified samples has the net effect of limiting undesirable intermodulation products, thereby enabling a coarser frequency resolution and/or lower degree of oversampling to be used. In one embodiment, the invention further includes a window function suitable for use with block-based cross-product enhanced HFR. A hardware embodiment of the invention may include an analysis filter bank, a subband processing unit configurable by control data and a synthesis filter bank.

Abstract: Techniques for receiving user input and soliciting responses from one or more users are described. A user may speak a question to a system. The system may determine that various question and answering content sources cannot provide an answer to the question. The system may then store data representing the question in a dedicated storage. The system may a topic, to which the question relates and provide the question to users having knowledge on the topic.

Abstract: Methods and systems for generation of a database schema compliant search query based on a natural language input are described herein. Natural language input may be received from a computing device. The natural language input may be associated with multiple search requests to a database. The natural language input may be parsed into a plurality of segments. The plurality of segments may be, for example, one or more words of a text string. At least one identifier for the plurality of segments may be associated with one or more confidence values. The natural language input may be converted into a single search query based on the confidence values and/or on a set of rules. The single search query may be initiated with respect to the database. The single search query may fetch content more efficiently than the multiple search requests.

Abstract: Systems and methods that access an online networked resource using a locator are disclosed. A first item of content on the networked resource is identified. A trigger rule comprising keywords and a sentiment classifier is accessed. A neural network, including input, hidden, and output layers, is used to assign a sentiment classification to the first item of content. The trigger rule, the sentiment classification, and identified keywords, are used to determine whether response content is to be posted to the online networked resource. In response to determining, using the trigger rule, the assigned sentiment classification, and keywords identified in the first item of content, that response content is to be posted to the online networked resource, the sentiment classification and identified keywords are used to select and/or generate a second item of content, and the second item of content is enabled to be posted to the online networked resource.

Abstract: Systems and methods for improving question-answering recall for complex, multi-sentence, convergent questions. More specifically, an autonomous agent accesses an initial answer that partly answers a question received from a user device. The agent represents the question and the initial answer as discourse trees. From the discourse trees, the agent identifies entities in the question that are not addressed by the answer. The agent forms an additional discourse tree from an additional resource such as a corpus of text. The additional discourse tree rhetorically connects a non-addressed entity with the answer. The agent designates this discourse tree as an imaginary discourse tree. When combined with the initial answer discourse tree, the imaginary discourse tree is used to generate an improved answer relative to existing solutions.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to facilitating predicting style breaches within content. In one embodiment, target content for which style breach prediction is desired is obtained. Style features associated with the target content are identified. Such style features and a style breach prediction model are used to predict a style breach within the target content, the style breach indicating a change of style used within the target content (e.g., a single document).

Abstract: In one embodiment, a method includes receiving a first user input by a first user at a client system associated with the first user, wherein the first user input is associated with a first language register, selecting a first language-register model from a plurality of language-register models based on the first language register by the client system, wherein the plurality of language-register models are each personalized with respect to the first user, and presenting a personalized first communication content generated based on the selected first language-register model at the client system.

Abstract: A mechanism is provided for implementing an intent segmentation mechanism that segments intent boundaries for multi-intent utterances in a conversational agent. For each term of a set of terms in the utterance from a real-time chat session, a set of adversarial utterances is generated for the utterance. An influence of changing each term is determined so as to identify a term importance value. Utilizing the term importance value, one or more of a change in ranking of the intent of the utterance or a change in confidence with regard to the intent of the utterance is identified. An entropy-based segmentation of the utterance into a plurality of candidate partitions is performed. An associated intent and entropy value are then assigned. Based on a segment with minimum entropy, a call associated with the real-time chat session is directed to an operation associated with an intent of the segment with minimum entropy.

Abstract: The present disclosure provides methods, devices and computer program products for non-uniform quantization of parameters. The disclosure further relates to a method and apparatus for reconstructing an audio object in an audio decoding system taking the non-uniformly quantized parameters into account. According to the disclosure, such an approach renders it possible to reduce bit consumption without substantially reducing the quality of the reconstructed audio object.

Abstract: A computer implemented method and system for processing an audio signal. The method includes the steps of extracting prosodic features from the audio signal, aligning the extracted prosodic features with a script derived from or associated with the audio signal, and segmenting the script with the aligned extracted prosodic features into structural blocks of a first type. The method may further include determining a distance measure between a structural block of a first type derived from the script with another structural block of the first type using, for example, the Damerau-Levenshtein distance.

Abstract: In one embodiment, a method includes receiving a first user input by a first user at a client system associated with the first user, wherein the first user input is associated with a first language register, selecting a first language-register model from a plurality of language-register models based on the first language register by the client system, wherein the plurality of language-register models are each personalized with respect to the first user, and presenting a personalized first communication content generated based on the selected first language-register model at the client system.

Abstract: To begin with, an acoustic model training apparatus extracts speech features representing speech characteristics, and calculates an acoustic-condition feature representing a feature of an acoustic condition of the speech data using an acoustic-condition calculation model that is represented as a neural network, based on an acoustic-condition calculation model parameter characterizing the acoustic-condition calculation model. The acoustic model training apparatus then generates an adjusted parameter that is an acoustic model parameter adjusted based on the acoustic-condition feature, the acoustic model parameter characterizing an acoustic model represented as a neural network to which an output layer of the acoustic-condition calculation model is coupled. The acoustic model training apparatus then updates the acoustic model parameter based on the adjusted parameter and the speech features, and updates the acoustic-condition calculation model parameters based on the adjusted parameter and the speech features.

Abstract: The disclosed method may include (1) determining whether a next operation of a plurality of operations of an artificial neural network (ANN) is dependent upon a Boolean predication value based on a representative value for a weight or an input of a node of the ANN, (2) based on the next operation not being dependent on the Boolean predication value, allowing the next operation to update a state of the ANN, and (3) based on the next operation being dependent on the Boolean predication value, performing at least one of (a) allowing, based on the Boolean predication value being a first value, the next operation to update the state of the ANN, and (b) preventing, based on the Boolean predication value being a second value different from the first value, the next operation from updating the state of the ANN. Various other methods and systems are also disclosed.

Abstract: A speech translation device includes: a beamformer which forms, from a speech signal obtained by a microphone array, a first beam and a second beam having different directions; a direction designator which designates one of the first beam and the second beam according to a user operation; a signal-to-noise (SN) ratio calculator which calculates an SN ratio using the designated beam as a signal component in the SN ratio, and the other beam not designated as a noise component; a display determiner which determines whether recognition of the designated beam is difficult, using the calculated SN ratio, and determines a speaking instruction for overcoming difficulty of the recognition when determining that the recognition is difficult; and a display which displays the speaking instruction determined by the display determiner in a display area.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer-readable media, for speech recognition using multi-dialect and multilingual models. In some implementations, audio data indicating audio characteristics of an utterance is received. Input features determined based on the audio data are provided to a speech recognition model that has been trained to output score indicating the likelihood of linguistic units for each of multiple different language or dialects. The speech recognition model can be one that has been trained using cluster adaptive training. Output that the speech recognition model generated in response to receiving the input features determined based on the audio data is received. A transcription of the utterance generated based on the output of the speech recognition model is provided.

Abstract: A computer system includes a processor configured to execute a conversational agent generator function. The conversational agent generator function is configured to receive a plurality of model input queries. Each model input query includes an intent and a corresponding answer to the intent. A database stores a corpus including sets of textual responses indicative of a common meaning. Each textual response in a given set of textual responses indicates a common meaning constructed in a different style indicative of a corresponding tone of communication. The processor is further configured to execute a rephrasing function configured determine the answer of one of the plurality of model input queries, and convert the answer of one of the plurality of model input queries into a plurality of different rephrased answers from the corpus stored in the database. The processor further outputs the plurality of different rephrased answers to the conversational agent generator function.

Abstract: A service, in response to receiving a question in a natural language format, identifies one or more selected passages from a corpus that are relevant to a focus of the question from among multiple passages in the corpus. The service aligns one or more answer grammatical properties of one or more answers, selected from the one or more selected passages, to one or more question grammatical properties of the focus of the question. The service returns the one or more answers in response to the question.

Abstract: The present technology relates to an information processing device, an information processing method, a voice output device, and a voice output method that allow a user to hear a speech from a home agent unit regardless of where the user is. The voice output device outputs, on the basis of an image and noise from a predetermined noise source which are obtained in a unit capable of outputting a voice toward a user, information indicating hearing difficulty of a voice from the unit at a position of the user. The present technology can be applied to a household voice assistant device that performs a speech to a user.