Abstract: Reference-sample feature vectors that quantify acoustic features of different respective portions of at least one reference speech sample, which was produced by a subject at a first time while a physiological state of the subject was known, are obtained. At least one test speech sample that was produced by the subject at a second time, while the physiological state of the subject was unknown, is received. Test-sample feature vectors that quantify the acoustic features of different respective portions of the test speech sample are computed. The test-sample feature vectors are mapped to respective ones of the reference-sample feature vectors, under predefined constraints, such that a total distance between the test-sample feature vectors and the respective ones of the reference-sample feature vectors is minimized. In response to the mapping, an output indicating the physiological state of the subject at the second time is generated. Other embodiments are also described.

Abstract: Aspects create a multimedia presentation wherein processors are configured to calculate a time it would take to narrate a plurality of words in a document at a specified speech speed in response to determining that the time it would take to narrate the plurality of words in the document at the specified speech speed exceeds a specified maximum time, generate a long summary of the document as a subset of the plurality of words, generate audio content for a first portion of the plurality of words of the long summary by applying a text-to-speech processing mechanism to the portion of the long summary at the desired speech speed, and create a multimedia slide of a multimedia presentation by adding the generated audio content to a presentation of text from a remainder portion of the plurality of words of the long summary.

Abstract: An Artificial Intelligence (AI) based document processing system receives a request including one or more documents related to a process to be automatically executed. The information including the fields and an intent required for the process are extracted from one or more of the request and the documents. The required documents and fields are selected based on the intent and a domain model. The required fields are validated using external knowledge and the discrepancies identified therein are resolved. An internal master document is built based on the required fields. The internal master document is employed for the automatic execution of the process which can include a de-identification process or an appeal process.

Abstract: Methods, systems, and apparatus including computer programs encoded on a computer storage medium, for generating a data set that associates each text segment in a vocabulary of text segments with a respective numeric embedding. In one aspect, a method includes providing, to an image search engine, a search query that includes the text segment; obtaining image search results that have been classified as being responsive to the search query by the image search engine, wherein each image search result identifies a respective image; for each image search result, processing the image identified by the image search result using a convolutional neural network, wherein the convolutional neural network has been trained to process the image to generate an image numeric embedding for the image; and generating a numeric embedding for the text segment from the image numeric embeddings for the images identified by the image search results.

Abstract: A system, a user terminal, a method of the system, a method of the user terminal, and a computer program product are provided. The system includes a communication interface, at least one processor operatively coupled to the communication interface, and at least one piece of memory operatively coupled to the at least one processor, wherein the at least one piece of memory is configured to store instructions configured for the at least one processor to receive sound data from a first external device through the communication interface, obtain a voice signal and a noise signal from the sound data using at least some of an automatic voice recognition module, change the voice signal into text data, determine a noise pattern based on at least some of the noise signal, and determine a domain using the text data and the noise pattern when the memory operates.

Abstract: The technology disclosed relates to authoring of vertical applications of natural language understanding (NLU), which analyze text or utterances and construct their meaning. In particular, it relates to new programming constructs and tools and data structures implementing those new applications.

Abstract: A dialogue system for a vehicle may include: an input processor configured to receive a user's utterance, to acquire an utterance text by recognizing the user's utterance, to recognize a dialogue subject based on the acquired utterance text, and to identify the user; and a dialogue manager including a memory storing program instructions and a processor configured to execute the stored program instructions, the dialogue manager configured to verify whether a chat room related to the dialogue subject is present, and to determine whether to add the identified user as a participant of the chat room based on a result of the verification.

Abstract: The present disclosure relates to template data processing, template data requesting, and template data presenting methods and devices, and a storage medium. The method includes obtaining a template subject and template key words matching the template subject and generating, according to the template key words, a first template message matching the template subject. The first template message includes the template key words. The method also includes generating a template identifier corresponding to the first template message and associating the template identifier with the corresponding first template message. The method further includes storing the template identifier and the corresponding first template message into a template library.

Abstract: Techniques are described herein for multi-modal interaction between users, automated assistants, and other computing services. In various implementations, a user may engage with the automated assistant in order to further engage with a third party computing service. In some implementations, the user may advance through dialog state machines associated with third party computing service using both verbal input modalities and input modalities other than verbal modalities, such as visual/tactile modalities.

Abstract: A method for determining a personality profile of an online user is disclosed. Social speech content data associated with an online user is stored. A machine learning model is used to determine a first personality profile of the online user based at least in part on the social speech content data associated with the online user. A second personality profile of the online user is determined based on the social speech content data using a scientific personality model encoded in an ontology, wherein the ontology encodes statistical relationships between a plurality of words and a plurality of personality traits based on one or more scientific research studies. An ensemble model is applied to determine a third personality profile of the online user based at least in part on the first personality profile and the second personality profile.

Abstract: Methods, systems, and computer readable medium for providing translated web content. A request is received from a user for content in a second language translated from content in a first language from a first Internet source. The content in the first language is obtained and divided into one or more translatable components. Whether the one or more translatable components have been previously translated, via at least one of machine translation, human translation, and a combination thereof, into the second language and stored as translated components in a storage is determined. If there are one or more translatable components previously translated and stored, the content is generated in the second language by modifying the content in the first language so that at least some translatable components are replaced with corresponding translated components and sent in the second language to the user as a response to the request.

Abstract: A speech recognition circuit comprises an input buffer for receiving processed speech parameters. A lexical memory contains lexical data for word recognition. The lexical data comprises a plurality of lexical tree data structures. Each lexical tree data structure comprises a model of words having common prefix components. An initial component of each lexical tree structure is unique. A plurality of lexical tree processors are connected in parallel to the input buffer for processing the speech parameters in parallel to perform parallel lexical tree processing for word recognition by accessing the lexical data in the lexical memory. A results memory is connected to the lexical tree processors for storing processing results from the lexical tree processors and lexical tree identifiers to identify lexical trees to be processed by the lexical tree processors.

Abstract: A keyword extraction method is provided.

Abstract: A method and apparatus for determining a speech interaction satisfaction are provided. The method includes: acquiring a speech interaction feature, the speech interaction feature including objective data of speech interaction and subjective data of the speech interaction, and the objective data and the subjective data being data for a given topic; evaluating the objective data to obtain an objective evaluation, and evaluating the subjective data to obtain a subjective evaluation; and obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model. Some embodiments may provide an authentic and comprehensive evaluation of the speech interaction.

Abstract: Embodiments of the present invention provide a speech recognition method and a terminal. The method includes: listening, by a speech wakeup apparatus, to speech information in a surrounding environment; when determining that the speech information obtained by listening matches a speech wakeup model, buffering, by the speech wakeup apparatus, speech information, of first preset duration, obtained by listening, and sending a trigger signal for triggering enabling of a speech recognition apparatus, where the trigger signal is used to instruct the speech recognition apparatus to read and recognize the speech information buffered by the speech wakeup apparatus; and recognizing first speech information buffered by the speech wakeup apparatus and the second speech information obtained by listening, to obtain a recognition result.

Abstract: An apparatus for determining a pitch information on the basis of an audio signal. The apparatus is configured to obtain a similarity value being associated with a given pair of portions of the audio signal having a given time shift, wherein the apparatus is configured to choose a length of signal portions of the audio signal used to obtain the similarity value for the given time shift in dependence on the given time shift and where the apparatus is configured to choose the length of the signal portions to be linearly dependent on the given time shift, within a tolerance of ±1 sample.

Abstract: The present disclosure relates to a method, apparatus, and system for speaker verification. The method includes: acquiring an audio recording; extracting speech signals from the audio recording; extracting features of the extracted speech signals; and determining whether the extracted speech signals represent speech by a predetermined speaker based on the extracted features and a speaker model trained with reference voice data of the predetermined speaker.

Abstract: An audio processor for processing an audio signal includes an audio signal phase measure calculator configured for calculating a phase measure of an audio signal for a time frame, a target phase measure determiner for determining a target phase measure for the time frame, and a phase corrector configured for correcting phases of the audio signal for the time frame using the calculated phase measure and the target phase measure to obtain a processed audio signal.

Abstract: A computing system is provided. The computing system includes a processor of a display device configured to execute one or more programs. The processor is configured to receive a command from a user by way of natural language input. The processor is configured to identity a set of candidate objects within or adjacent a user's field of view having associated spatialized regions on which the command can be executed, the set of candidate objects identified at least partially by using a machine learning model. The processor is configured to use visual or audio indicators associated with the candidate objects and query the user for disambiguation input. The processor is configured to receive the disambiguation input from the user that selects a target object, executing the command on the target object. The processor is configured to train the machine learning model using the disambiguation input and data about the spatialized regions.

Abstract: System and methods for controlling healthcare devices and systems using voice commands are presented. In some aspects a listening device may receive voice command from a person. The voice command may be translated into human readable or machine readable text via a speech-to-text service. A control component may receive the text and send device-specific instructions to a medical device associated with a patient based on the translated voice command. In response to the instructions, the medical device may take an action on a patient. Some examples of actions taken may include setting an alarm limit on a monitor actively monitoring a patient and adjusting the amount of medication delivered by an infusion pump. Because these devices may be controlled using a voice command, in some cases, no physical or manual interaction is needed with the device. As such, multiple devices may be hands-free controlled from any location.