Abstract: A system and method for analyzing sound signals within a predetermined space, including: analyzing a plurality of sound signals captured within a predetermined space via at least one sound sensor; generating a grid corresponding to the predetermined space based on the plurality of sound signals, wherein the grid is utilized to identify areas within the predetermined space as interest points; identifying, based on the interest point, at least one sound generating object within the grid based on the analysis of the plurality of sound signals; and identifying at least one characteristic of the plurality of sound signals.

Abstract: Provided are an electronic device and speech recognition method therefor. The electronic device may include a communication interface to receive speech data from an external electronic device, a memory to store a common language model used by default for speech recognition, a first language model designated for each user, a second language model associated with context information of each user, and a third language model associated with words collected by the electronic device for a preset period of time from the reception time of the speech data; and a processor to perform a procedure of combining at least one of the first language model, the second language model, and the third language model with the common language model to construct an integrated language model, performing speech recognition on the basis of the speech data and the integrated language model, and outputting a speech recognition result corresponding to the speech data.

Abstract: A method of audio processing comprises receiving an audio signal. A plurality of framed versions of the received audio signal are formed, each of the framed versions having a respective frame start position. One of the plurality of framed versions of the received audio signal is selected. The selected one of the plurality of framed versions of the received audio signal is used in a subsequent process.

Abstract: A method comprising: operating at least one hardware processor for: receiving, as input, at least one named entity, modifying said named entity based on a plurality of modification rules to generate a set of candidate named entities corresponding to said named entity, and identifying, for at least one candidate named entity in said set of candidate named entities, an article in a knowledge base of articles, wherein a title of said article matches said candidate named entity.

Abstract: A spectrum encoding method includes selecting an important spectral component in band units for a normalized spectrum and encoding information of the selected important spectral component for a band, based on a number, a position, a magnitude and a sign thereof. A spectrum decoding method includes obtaining from a bitstream, information about an important spectral component for a band of an encoded spectrum and decoding the obtained information of the important spectral component, based on a number, a position, a magnitude and a sign of the important spectral component.

Abstract: Text analysis, specifically, narratives, wherein identification of distinct and independent participants (entities of interest) in a narrative is an important task for many NLP applications. This task becomes challenging because these participants are often referred to using multiple aliases. Identifying aliases of participants in a narrative is crucial for NLP applications. Existing conventional methods are supervised for alias identification which requires a large amount of manually annotated (labeled) data and are also prone to errors. Embodiments of the present disclosure provide systems and methods that implement Markov Logic Network (MLN) to encode linguistic knowledge into rules for identification of aliases for aliases mention identification using proper nouns, pronouns or noun phrases with common noun headword.

Abstract: Systems and methods to generate a cognitive model are described. A particular example of a system includes a memory including program code having an application programming interface and a user interface, and a processor configured to access the memory and to execute the program code to generate a cognitive model, to run analysis on the cognitive model to determine a factor that is impacting a performance of the cognitive model, to determine an action based on the factor, to report at least one of the factor and the action to a user, and to use the action to generate a second cognitive model.

Abstract: A dialogue system, comprises an input unit configured to acquire utterance contents of a user in a dialogue; a mode determining unit configured to determine, based on the utterance contents acquired by the input unit, whether a mode of the dialogue is task-oriented or non-task-oriented; a plurality of intention understanding units each corresponding to a specific domain; and a domain determining unit configured to determine, when the mode of the dialogue is task-oriented, a domain of the dialogue based on a result of intention understanding of the utterance contents performed using each of the plurality of intention understanding units.

Abstract: A method for voice processing includes acquiring sound information, extracting speech information from the sound information, recognizing semantic information of the speech information, obtaining context information, and determining response information based on the semantic information and the context information.

Abstract: A control method for allowing a user to specify an electronic device and switch it to a speech recognition mode is provided. With the optimum control method and the electronic device utilizing the method, a voice command may be transmitted to the electronic device more quickly and effectively regardless of the circumstances, and the electronic device may be specified through gesture recognition to enable transmission of the voice command, so that the voice command may be effectively executed without needing a user to learn or memorize a name or the like of the electronic device in advance for speech recognition. Further, it is possible to more accurately recognize a gesture that is a preliminary step for transmitting a voice command to the electronic device, thereby improving the recognition rate and preventing malfunction.

Abstract: A method of training word embeddings is provided. The method includes determining anchors, each comprising a first word in a first domain and a second word in a second domain, training word embeddings for the first and second domains, and training a transform for transforming word embedding vectors in the first domain to word embedding vectors in the second domain, wherein the training minimizes a loss function that includes an anchor loss for each anchor, such that for each anchor, the anchor loss is based on a distance between the anchor's second word's embedding vector and the transform of the anchor's first word's embedding vector, and for each anchor, the anchor loss for the respective anchor is zero when the distance between the respective anchor's second word's embedding vector and the transform of the respective anchor's first word's embedding vector is less than a specific tolerance.

Abstract: In one example, an apparatus includes: a wavelet transform engine to receive a first signal stream and perform a wavelet transform on a first time domain sample of the first signal stream, the first wavelet transform engine to output at least one first coefficient for a first frequency range; an energy calculation circuit to compute a first energy signature for the at least one first coefficient; and a correlation circuit to generate a correlation value using the first energy signature, a second energy signature and a plurality of previous energy signatures.

Abstract: Devices and techniques are generally described for shared execution caching of contextual data. A first request to retrieve a first value from a first field of a database may be received. First entity identifier data associated with the first request may be determined. A second request to retrieve the first value from the first field may be received, the second request being associated with second entity identifier data. A determination may be made that a first host device is associated with the first entity identifier data and the second entity identifier data. In some examples, a first executable instruction effective to retrieve the first value from the first field of the database when executed by the first host device may be generated. In some examples, the first value may be retrieved from the first field of the database and sent to the first speech processing component.

Abstract: A computer-implemented method for supporting combinations of intents in a conversation service. In one embodiment, the computer-implemented method provides a conversation service using a set of conversation trees. Each conversation tree within the set of conversation trees comprises a list of intents that define how to answer to a certain type of user input. The computer-implemented method creates an intent graph using the list of intents of each conversation tree within the set of conversation trees. The computer-implemented method adds composite intents to the set of conversation trees using the intent graph to generate a set of updated conversation trees. The computer-implemented method configures the conversation service to use the set of updated conversation trees that comprises composite intents for supporting combinations of intents.

Abstract: A method includes: receiving time instants of audio signals generated by a set of microphones at a location; determining a distortion measure between frequency components of at least some of the received audio signals; determining a similarity measure for the frequency components using the determined distortion measure; and processing the audio signals based on the determined similarity measure.

Abstract: Methods, apparatuses, and computer program products are described herein that are configured to generate a situational analysis text. In some example embodiments, a method is provided that comprises generating a set of messages based on one or more key events in a primary data channel and one or more significant events in one or more related data channels in response to an alert condition. The method of this embodiment may also include generating a situational analysis text based on the set of messages and the relationships between them. In some example embodiments, the situational analysis text is configured to linguistically express the one or more key events, the one or more significant events, and the relationships between the one or more key events and the one or more significant events.

Abstract: A method, which is performed in an electronic device, for activating a target application is disclosed. The method may include receiving an input sound stream including an activation keyword for activating the target application and a speech command indicative of a function of the target application. The method may also detect the activation keyword from the input sound stream. If the activation keyword is detected, a portion of the input sound stream including at least a portion of the speech command may be buffered in a buffer memory. In addition, in response to detecting the activation keyword, the target application may be activated to perform the function of the target application.

Abstract: A computer-implemented method for producing healthcare data records from graphical inputs by computer users includes receiving, on a graphical user interface of a computer system, a user identification of a diagnosis for a patient, the user identification produced by user selection on the graphical user interface; identifying one or more parameters that characterize the diagnosis; displaying on the graphical user interface a plurality of selectable values for particular ones of identified parameters; receiving sequential user selections representations of particulars ones of the values; and generating an electronic medical record representation that represents the identified diagnosis having the selected values for the one or more parameters.

Abstract: Extracting one or more keywords in a claim, including: decomposing an independent claim into a plurality of elements, wherein each element in the plurality of elements has at least one term; constructing a dependency structure from the plurality of elements, wherein each element in the plurality of elements has a depth in the dependency structure; calculating a score using the depth of the element corresponding to the at least one term in the independent claim; and extracting at least one keyword from the at least one term having a score equal to or larger than a predetermined threshold.

Abstract: A system configured to perform directional speech separation. The system may dynamically associate direction-of-arrivals with one or more audio sources in order to generate output audio data that separates each of the audio sources. The system identifies a target direction for each audio source, dynamically determines directions that are correlated with the target direction, and generates output signals for each audio source. The system may associate individual frequency bands with specific directions based on a time delay detected by two or more microphones. The system may determine a cross-correlation between each direction and the target direction and select directions with strong correlation. The system may generate time-frequency mask data indicating frequency bands corresponding to the directions associated with a particular audio source. Using the mask data, the system generates output audio data specific to the audio source, resulting in directional speech separation between different audio sources.