Abstract: An automatic speech recognition with detection of at least one contextual element, and application to aircraft flying and maintenance are provided. The automatic speech recognition device comprises a unit for acquiring an audio signal, a device for detecting the state of at least one contextual element, and a language decoder for determining an oral instruction corresponding to the audio signal. The language decoder comprises at least one acoustic model defining an acoustic probability law and at least two syntax models each defining a syntax probability law.

Abstract: A method includes storing a meaning taxonomy including meaning loaded entities and associations between the meaning loaded entities and a plurality of syntactic structures; receiving, from a first source, first content including one or more first syntactic structures; identifying, based at least in part on identification criteria, one or more meaning loaded entities that are linked to the one or more first syntactic structures by one or more first associations of the plurality of associations, the identification criteria including at least one of a type of the first content, a document size of the first content, a publication date of the first content, and an importance of the meaning load entities; calculating a first content summary indicating a level of coverage of the meaning loaded entities within the first content; and generating a first visual representation of the first content summary.

Abstract: Certain example embodiments relate to speech privacy systems and/or associated methods. The techniques described herein disrupt the intelligibility of the perceived speech by, for example, superimposing onto an original speech signal a masking replica of the original speech signal in which portions of it are smeared by a time delay and/or amplitude adjustment, with the time delays and/or amplitude adjustments oscillating over time. In certain example embodiments, smearing of the original signal may be generated in frequency ranges corresponding to formants, consonant sounds, phonemes, and/or other related or non-related information-carrying building blocks of speech. Additionally, or in the alternative, annoying reverberations particular to a room or area in low frequency ranges may be “cut out” of the replica signal, without increasing or substantially increasing perceived loudness.

Abstract: Systems and methods for multiple topic automated chatting are provided. The systems and method provide multiple topic automated (or artificial intelligence) chatting by analyzing user inputs in a conversation to determine a plurality topics, to determine and score features related to the determined topics and different users, and to create a knowledge graph of the determined topics. Based on these determinations, the systems and methods may determine if a reply should be provided and then predict a reply.

Abstract: An information processing system to increase weights of words that are related to a text, but that do not explicitly occur in the text, in a weight vector representing the text, is provided. An adjusting system (100) includes a distance storing unit (110) and an adjusting unit (120). The distance storing unit (110) stores distances between any two terms of a plurality of terms. The distance between two terms becomes smaller as the two terms are semantically more similar. The adjusting unit (120) adjusts a weight of each term of the plurality of terms in a weight vector including weights of the plurality of terms and representing a text, on the basis of a distance between each term and other term in the weight vector and a weight of the other term.

Abstract: A method for speech processing includes determining a first unvoicing parameter for a first subframe of a speech signal, and determining a smoothed unvoicing parameter for the first subframe according to a second unvoicing parameter of a second subframe prior to the first subframe of the speech signal. The first unvoicing parameter is determined according to a periodicity parameter and a spectral tilt parameter. The method further includes computing a difference between the first unvoicing parameter for the first subframe and the smoothed unvoicing parameter for the first subframe and determining a classification of the first subframe using the computed difference as a decision parameter. The classification indicates whether the first subframe is an unvoiced speech signal or not an unvoiced speech signal. Bandwidth extension is performed on the speech signal for the first subframe according to the classification of the first subframe.

Abstract: Embodiments of end-to-end deep learning systems and methods are disclosed to recognize speech of vastly different languages, such as English or Mandarin Chinese. In embodiments, the entire pipelines of hand-engineered components are replaced with neural networks, and the end-to-end learning allows handling a diverse variety of speech including noisy environments, accents, and different languages. Using a trained embodiment and an embodiment of a batch dispatch technique with GPUs in a data center, an end-to-end deep learning system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale.

Abstract: Embodiments of end-to-end deep learning systems and methods are disclosed to recognize speech of vastly different languages, such as English or Mandarin Chinese. In embodiments, the entire pipelines of hand-engineered components are replaced with neural networks, and the end-to-end learning allows handling a diverse variety of speech including noisy environments, accents, and different languages. Using a trained embodiment and an embodiment of a batch dispatch technique with GPUs in a data center, an end-to-end deep learning system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale.

Abstract: The invention relates to a method for automatically detecting meaning patterns in a text using a plurality of input words, in particular a text with at least one sentence, comprising a database system containing words of a language, a plurality of defined categories of meaning in order to describe the properties of the words, and meaning signals for all the words stored in the database, wherein a meaning signal is a clear numerical characterization of the meaning of the words using the categories of meaning.

Abstract: An audio signal (X) is represented by a bitstream (B) segmented into frames. An audio processing system (500) comprises a buffer (510) and a decoding section (520). The buffer joins sets of audio data (D1; D2, . . . , DN) carried by N respective frames (F1, F2, . . . , FN) into one decodable set of audio data (D) corresponding to a first frame rate and to a first number of samples of the audio signal per frame. The frames have a second frame rate corresponding to a second number of samples of the audio signal per frame. The first number of samples is N times the second number of samples. The decoding section decodes the decodable set of audio data into a segment of the audio signal by at least employing signal synthesis, based on the decodable set of audio data, with a stride corresponding to the first number of samples of the audio signal.

Abstract: Systems and methods for determining confidence levels associated with attribute values of informational objects. An example method comprises: receiving a natural language text; performing syntactico-semantic analysis of the natural language text to produce a plurality of semantic structures; interpreting the plurality of semantic structures using a set of production rules to produce a plurality of data items, each data item associating an attribute value with an informational object representing an entity referenced by the natural language text; and determining, for at least one data item of the plurality of data items, a confidence level associated with the attribute value, by evaluating a confidence function associated with the set of production rules.

Abstract: Certain example embodiments relate to speech privacy systems and/or associated methods. The techniques described herein disrupt the intelligibility of the perceived speech by, for example, superimposing onto an original speech signal a masking replica of the original speech signal in which portions of it are smeared by a time delay and/or amplitude adjustment, with the time delays and/or amplitude adjustments oscillating over time. In certain example embodiments, smearing of the original signal may be generated in frequency ranges corresponding to formants, consonant sounds, phonemes, and/or other related or non-related information-carrying building blocks of speech. Additionally, or in the alternative, annoying reverberations particular to a room or area in low frequency ranges may be “cut out” of the replica signal, without increasing or substantially increasing perceived loudness.

Abstract: A separation unit separates voice signals of a plurality of channels into an incoming component in each incoming direction, a selection unit selects a statistic corresponding to an incoming direction of the incoming component separated by the separation unit from a storage unit which stores a predetermined statistic and a voice recognition model for each incoming direction, an updating unit updates the voice recognition model on the basis of the statistic selected by the selection unit, and a voice recognition unit recognizes a voice of the incoming component separated using the voice recognition model.

Abstract: Provided is a process of modifying semantic similarity graphs representative of pair-wise similarity between documents in a corpus, the method comprising obtaining a semantic similarity graph that comprises more than 500 nodes and more than 1000 weighted edges, each node representing a document of a corpus, and each edge weight indicating an amount of similarity between a pair of documents corresponding to the respective nodes connected by the respective edge; obtaining an n-gram indicating that edge weights affected by the n-gram are to be increased or decreased; expanding the n-gram to produce a set of expansion n-grams; adjusting edge weights of edges between pairs of documents in which members of the expanded n-gram set co-occur.

Abstract: An audio encoding apparatus and method that encodes hybrid contents including an object sound, a background sound, and metadata, and an audio decoding apparatus and method that decodes the encoded hybrid contents are provided. The audio encoding apparatus may include a mixing unit to generate an intermediate channel signal by mixing a background sound and an object sound, a matrix information encoding unit to encode matrix information used for the mixing, an audio encoding unit to encode the intermediate channel signal, and a metadata encoding unit to encode metadata including control information of the object sound.

Abstract: A speech recognition system and method thereof, a vocabulary establishing method and a computer program product are provided.

Abstract: A multilingual power system protection device to facilitate communications in different human languages over different communication ports is described herein. In one embodiment, an electric power system device may include communication ports configured to receive inputs in different human languages and a monitored equipment interface in communication with a component of an electric power system. The electric power system device may receive an input on one of the communication ports and may associate the input with a function implemented by the component of the electric power system. The electric power system device may further generate a response in a pre-selected human language, and may transmit the response using the communication ports.

Abstract: Methods and systems are provided for communicating an announcement to passengers on a transportation vehicle. For example, one method includes providing an information system on the vehicle having at least one of a wireless access point and a plurality of seat display devices and operating the information system to communicate with the wireless access point or the seat display devices. The method includes playing audio corresponding to the announcement over a public address system of the vehicle, and causing text corresponding to the audio to display on the seat display devices or personal electronic devices in communication with the wireless access point.

Abstract: The present disclosure relates to a portable terminal, and more particularly, to a portable terminal and a method of detecting a gesture and controlling a function. A method of controlling a function of a portable terminal includes: detecting a gesture; activating a voice recognition module in response to the detected gesture; and analyzing a voice input into the activated voice recognition module, and executing a function corresponding to the input voice.

Abstract: One embodiment of the present invention provides a system for facilitating information in multiple languages based on an optical code. During operation, the system scans an optical code accompanying a text phrase and receives a location of a conversion server embedded in the optical code. The system then determines one or more target languages for obtaining the text phrase in the target languages and sends a query message to the conversion server based on the retrieved location. The query message comprises a list of the one or more target languages.