Abstract: A computer-implemented method for constructing an ontology includes using a chatbot program to elicit an utterance from a user. An utterance is received from the user. The received utterance is parsed to detect a concept/property/relationship therein. It is determined whether the identified concept/property/relationship is represented within an ontology and the identified concept/property/relationship is added to the ontology when it is determined that the it is not present therein. A response to the received utterance is generated using the ontology. The generated response is presented to the user using the chatbot.

Abstract: Techniques for implementing a “volatile” user ID are described. A system receives first input audio data and determines first speech processing results therefrom. The system also determines a first user that spoke an utterance represented in the first input audio data. The system establishes a multi-turn dialog session with a first content source and receives first output data from the first content source based on the first speech processing results and the first user. The system causes a device to present first output content associated with the first output data. The system then receives second input audio data and determines second speech processing results therefrom. The system also determines the second input audio data corresponds to the same multi-turn dialog session. The system determines a second user that spoke an utterance represented in the second input audio data and receives second output data from the first content source based on the second speech processing results and the second user.

Abstract: A method, computer system, and a computer program product for adaptively selecting an acoustic feature extractor in an Artificial Intelligence system is provided. The present invention may include acquiring a frame of an acoustic signal. The present invention may include checking a status of a flag to be used to indicate a proper acoustic feature extractor to be selected. The present invention may include processing the frame of the acoustic signal by the selected acoustic feature extractor indicated by the checked status. The present invention may include determining, based on data generated in the processing of the frame of the acoustic signal, an actual status of the frame of the acoustic signal. The present invention may include updating the status of the flag according to the actual status.

Abstract: Techniques for communicating announcements across devices are described. A system performs speech processing on input audio data to determine a command to output an announcement. The system determines devices outputting indicators representing users are proximate thereto. The system generates output audio data corresponding to at least a portion of the input audio data corresponding to announcement content. The system sends the output audio data to the devices along with instructions to output audio corresponding to the output audio data.

Abstract: Methods, apparatus, systems and articles of manufacture to segment audio and determine audio segment similarities are disclosed. A disclosed example method includes developing features characterizing audio with a neural network, computing a self-similarity matrix based on the features, and identifying segments of the audio based on the self-similarity matrix.

Abstract: A device includes a processor configured to determine a feature vector based on an utterance and to determine a first embedding vector by processing the feature vector using a trained embedding network. The processor is configured to determine a first distance metric based on distances between the first embedding vector and each embedding vector of a speaker template. The processor is configured to determine, based on the first distance metric, that the utterance is verified to be from a particular user. The processor is configured to, based on a comparison of a first particular distance metric associated with the first embedding vector to a second distance metric associated with a first test embedding vector of the speaker template, generate an updated speaker template by adding the first embedding vector as a second test embedding vector and removing the first test embedding vector from test embedding vectors of the speaker template.

Abstract: A translation device includes a character string output unit, a fixed phrase acquisition unit, a matching unit, a division processor, and a translation result output unit. The character string output unit outputs a plurality of character strings each configuring at least part of text in a first language and including a head character of the text, in order of character string length. The fixed phrase acquisition unit acquires a fixed phrase in the first language associated with a translation phrase in a second language. The matching unit calculates a similarity between each of the character strings and the fixed phrase in order of the character string length. The division processor divides the text at a position achieving a maximum similarity among values of the similarity exceeding a first threshold. The translation result output unit outputs the translation phrase as translation of the text thus divided.

Abstract: The present disclosure provides a method and a device for recognizing a text segmentation position. The method includes: receiving a continuous voice message inputted by a user, and recognizing the continuous voice message to generate a text message corresponding to the continuous voice message; analyzing the text message to determine an interval position, and sequentially inserting a sentence end and sentence begin sign at each interval position; calculating a segmentation value corresponding to the sentence end and sentence begin sign inserted at a present interval position according to a preset algorithm; and determining whether the segmentation value is greater than a preset threshold, and determining the present interval position as a segmentation position when the segmentation value is greater than the preset threshold.

Abstract: Embodiments of the invention relate to an error concealment unit for providing an error concealment audio information for concealing a loss of an audio frame in an encoded audio information. The error concealment unit provides a first error concealment audio information component for a first frequency range using a frequency domain concealment. The error concealment unit also provides a second error concealment audio information component for a second frequency range, which includes lower frequencies than the first frequency range, using a time domain concealment. The error concealment unit also combines the first error concealment audio information component and the second error concealment audio information component, to obtain the error concealment audio information. Other embodiments of the invention relate to a decoder including the error concealment unit, as well as related encoders, methods, and computer programs for decoding and/or concealing.

Abstract: A generally diverse set of audiovisual clips is sourced from one or more repositories for use in preparing a coordinated audiovisual work. In some cases, audiovisual clips are retrieved using tags such as user-assigned hashtags or metadata. Pre-existing associations of such tags can be used as hints that certain audiovisual clips are likely to share correspondence with an audio signal encoding of a particular song or other audio baseline. Clips are evaluated for computationally determined correspondence with an audio baseline track. In general, comparisons of audio power spectra, of rhythmic features, tempo, pitch sequences and other extracted audio features may be used to establish correspondence. For clips exhibiting a desired level of correspondence, computationally determined temporal alignments of individual clips with the baseline audio track are used to prepare a coordinated audiovisual work that mixes the selected audiovisual clips with the audio track.

Abstract: A lyrics analyzer generates tags and explicitness indicators for a set of tracks. These tags may indicate the genre, mood, occasion, or other features of each track. The lyrics analyzer does so by generating an n-dimensional vector relating to a set of topics extracted from the lyrics and then using those vectors to train a classifier to determine whether each tag applies to each track. The lyrics analyzer may also generate playlists for a user based on a single seed song by comparing the lyrics vector or the lyrics and acoustics vectors of the seed song to other songs to select songs that closely match the seed song. Such a playlist generator may also take into account the tags generated for each track.

Abstract: Systems and processes for proactive assistance based on dialog communication between devices are provided. In one example process, while voice communication between an electronic device and a second electronic device is established, a stream of audio data associated with the second electronic device can be received. In response to detecting a user input, a text representation of speech contained in a portion of the stream of audio data can be generated. The process can determine whether the text representation contains information corresponding to one of a plurality of types of information. In response to determining that the text representation contains information corresponding to one of a plurality of types of information, one or more tasks based on the information can be performed.

Abstract: The present disclosure involves systems, software, and computer implemented methods for personalizing interactions within a conversational interface based on an input context. One example system performs operations including receiving a conversational input received via a conversational interface. The conversational input is analyzed to determine an intent and lexical personality score based on the input's characteristics. A set of responsive content is determined and includes a set of initial tokens representing an initial response. A set of synonym tokens associated with at least some of the initial tokens are identified, and at least one synonym token associated with a similar lexical personality score to the input is determined. At least one of the initial tokens are replaced with the determined synonym token to generate a modified version of the set of response content. The modified version of the response is then transmitted to a device in response to the input.

Abstract: A controller for classifying devices of a building management system (BMS). The controller may be configured to obtain an entity name for a device, extract a core name from the entity name, compare the core name to candidate core names, determine scores for each comparison, identify a highest score, identify a class of a candidate core name, and classify the device in the class.

Abstract: The present disclosure provides a dialogue system and a method for controlling thereof. The dialogue system may include: an input processor configured to authenticate a user and receive new input information and new state information of the user; a storage configured to store existing state information of the user, existing input information, and available services; a controller configured to identify a service based on the new input information and the existing input information, and to identify the service based on the new state information of the user and the existing state information of the user, wherein the service is configured to fit needs of the user; and an output processor configured to determine a service format based on the new input information and the new state information of the user, wherein the service format is regarding ways to provide the service to the user.

Abstract: A computer-implemented method determines a category of a request provided by a user by means of a user device. The user device includes connection means and means for receiving a request description relating to said request from said user. The method includes receiving, from the user, the request description, by means of the device, and uploading the request description to a server. The server has access to a database which includes a number of previously categorized requests each including a category and a vocabulary, which includes a number of word vector representations. The method further includes identifying, by the server, a number of component words belonging to a natural language text string included in the request description; obtaining, for at least one of the component words, an associated word vector representation from the vocabulary, and determining a request vector, based on at least one obtained word vector representation.

Abstract: The present disclosure provides a method for supporting translation of global languages, and the product thereof. The method includes the following steps: receiving, by a smart phone, a calling request sent by a terminal, connecting the calling request, and establishing a calling connection; receiving, by the smart phone, first voice information transmitted through the calling connection, identifying a first language and a first dialect that correspond to the first voice information, obtaining a translation model corresponding to the first dialect, and translating the first voice information of the first dialect into second voice information of a second dialect; and playing, by the smart phone, the second voice information of the second dialect by using a speaker device.

Abstract: A method for spectrum recovery in spectral decoding of an audio signal, comprises obtaining of an initial set of spectral coefficients representing the audio signal, and determining a transition frequency. The transition frequency is adapted to a spectral content of the audio signal. Spectral holes in the initial set of spectral coefficients below the transition frequency are noise filled and the initial set of spectral coefficients are bandwidth extended above the transition frequency. Decoders and encoders being arranged for performing part of or the entire method are also illustrated.

Abstract: A received signal represents a user's speech. A first speaker recognition process is performed on a first portion of the received signal, to obtain a first output result. A second speaker recognition process is performed on a second portion of the received signal that is different from the first portion of the received signal, to obtain a second output result. The second speaker recognition process is different from the first speaker recognition process. The first and second output results are combined to obtain a combined output result indicating a likelihood that the user is a registered user.

Abstract: A system and method is provided for providing searchable customer call indexes. Consistent with disclosed embodiments, a system may receive call information associated with telephone conversations between callers and a vendor, the call information including an audio recording or transcript for each telephone conversation. The system may also identify one or more keywords from the audio recordings or transcripts and index the call information into one or more indexes based on the identified keywords. Finally, the system may determine search results responsive to a search query based on the indexing. In some embodiments, changes to customer service may be identified based on the search results.