Abstract: An example apparatus for detecting keywords in audio includes an audio receiver to receive audio comprising a keyword to be detected. The apparatus also includes a spike transducer to convert the audio into a plurality of spikes. The apparatus further includes a spiking neural network to receive one or more of the spikes and generate a spike corresponding to a detected keyword.

Abstract: Embodiments are directed to a natural language processing (NLP) system configured to receive a natural language (NL) input and perform an analysis operation to generate a NL output. The NLP system is configured to generate at least one confidence level based at least in part on at least one portion of the analysis operation. The NLP system is further configured to integrate at least one disfluency into the NL output based at least in part on the at least one confidence level.

Abstract: Identification of keywords from media content metadata including caption data is provided. When a piece of media content is received by a user, media content metadata and caption data may be provided to a data mining engine operable to identify and tag keywords. Identified keyword data may be provided to one or more native or third party applications for providing information to the user relevant to what he is watching.

Abstract: A method and system is provided for generating a group of phonetically similar masked data. The present application provides a method and system for generating a group of phonetically similar masked data; comprises preprocessing of input dataset values comprising a list of fictitious data values to be used as masked data; determining a plurality of groups of phonetically similar data values present in the dataset list; and deriving metaphone for each input data value to be masked; generating a first numeric code from derived metaphone value of input data value to be masked; selecting one group of phonetically similar data values out of the plurality of groups of phonetically similar data values based on the generated first numeric code; and generating a second numeric code from input data value for selecting a masked value from a plurality of fictitious data group.

Abstract: Embodiments are directed to a natural language processing (NLP) system configured to receive a natural language (NL) input and perform an analysis operation to generate a NL output. The NLP system is configured to generate at least one confidence level based at least in part on at least one portion of the analysis operation. The NLP system is further configured to integrate at least one disfluency into the NL output based at least in part on the at least one confidence level.

Abstract: A computer-implemented method can include receiving, at a computing device having one or more processors, a user input in a first writing system. The method can include determining one or more candidates based on the user input that correspond to the user input represented in a second writing system different than the first writing system. The method can include determining whether at least one of the candidates corresponds to a proper name of a geographic location and, when at least one of the candidates corresponds to the proper name of the geographic location, determining a local name of the geographic location. The local name can be the proper name of the geographic location in a local language associated with the geographic location. The method can include augmenting the one or more candidates to include the local name to obtain an augmented candidate list, and outputting the augmented candidate list.

Abstract: Compact finite state transducers (FSTs) for automatic speech recognition (ASR). An HCLG FST and/or G FST may be compacted at training time to reduce the size of the FST to be used at runtime. The compact FSTs may be significantly smaller (e.g., 50% smaller) in terms of memory size, thus reducing the use of computing resources at runtime to operate the FSTs. The individual arcs and states of each FST may be compacted by binning individual weights, thus reducing the number of bits needed for each weight. Further, certain fields such as a next state ID may be left out of a compact FST if an estimation technique can be used to reproduce the next state at runtime. During runtime portions of the FSTs may be decompressed for processing by an ASR engine.

Abstract: An input system includes a first gesture detection unit and a second gesture detection unit. The first gesture detection unit includes a first light emitting device for emitting a first light beam, a first light sensing device for receiving the first light beam reflected by a first motion trajectory generated by a user and outputting a first image signal, and a first processing unit for processing the first image signal and outputting a first command signal. The second gesture detection unit includes a second light emitting device for emitting a second light beam, a second light sensing device for receiving the second light beam reflected by a second motion trajectory generated by the user and outputting a second image signal, and a second processing unit for processing the second image signal and outputting a second command signal.

Abstract: A technique improves training and speech quality of a text-to-speech (TTS) system having an artificial intelligence, such as a neural network. The TTS system is organized as a front-end subsystem and a back-end subsystem. The front-end subsystem is configured to provide analysis and conversion of text into input vectors, each having at least a base frequency, f0, a phenome duration, and a phoneme sequence that is processed by a signal generation unit of the back-end subsystem. The signal generation unit includes the neural network interacting with a pre-existing knowledgebase of phenomes to generate audible speech from the input vectors. The technique applies an error signal from the neural network to correct imperfections of the pre-existing knowledgebase of phenomes to generate audible speech signals. A back-end training system is configured to train the signal generation unit by applying psychoacoustic principles to improve quality of the generated audible speech signals.

Abstract: The present invention relates to a method, an apparatus, and a computer-readable recording medium for improving at least one semantic unit set by using a phonetic sound. The method for improving a set including at least one semantic unit, in which the set including at least one semantic unit is a captured semantic unit set, includes: receiving an improvement phonetic sound according to utterance of a user; specifying an improvement semantic unit set on the basis of the improvement phonetic sound; specifying a semantic unit set as a target to be actually improved within the captured semantic unit set, as a matched semantic unit set, on the basis of correlation thereof with the improvement semantic unit set; and replacing the matched semantic unit set within the captured semantic unit set with the improvement semantic unit set.

Abstract: Systems and methods for controlling related video content based on domain specific language models are disclosed. Some embodiments include one or more processors configured to obtain a first audio information of first video content, identify first primary time codes based on the first audio information, obtain second audio information of second video content and identify a first secondary plurality of time codes based on the second audio information, determine an amount of differences between the first primary time codes and first secondary time codes, and generate information indicating whether the first video content and the second video content are related based on the determined amount of differences.

Abstract: The present invention extends to methods, systems, and computer program products for deriving signal location from signal content. A raw signal is ingested. A partially normalized signal is derived from the raw signal. One or more named entities are recognized from content included in the raw signal. A geo cell database is queried with the one or more recognized named entities. A geo cell where the raw signal originated is identified. The partially normalized signal is annotated with a location annotation identifying the geo cell. A location is determined from the location annotation. The location is inserted into the partially normalized signal to form a normalized signal.

Abstract: The present disclosure provides a voiceprint authentication method and a voiceprint authentication apparatus. The method includes: displaying a tip text to a user, the tip text being a combination of a preregistered phrase; obtaining a speech of the tip text read by the user; obtaining a pre-established registration model and determining a result of a voiceprint authentication according to the speech of the tip text and the pre-established registration model, if the speech of the tip text corresponds to the tip text.

Abstract: A speech interface device may capture user speech for analysis by automatic speech recognition (ASR) and natural language understanding (NLU) components. However, an audio signal representing the user speech may also contain interfering sound generated by a media player that is playing audio content such as music. Before performing ASR and NLU, a system attempts to identify the content being played by the media player, such as by querying the media player or by analyzing the audio signal. The system then obtains the same content from an available source and subtracts the audio represented by the content from the audio signal.

Abstract: A technique improves training and speech quality of a text-to-speech (TTS) system having an artificial intelligence, such as a neural network. The TTS system is organized as a front-end subsystem and a back-end subsystem. The front-end subsystem is configured to provide analysis and conversion of text into input vectors, each having at least a base frequency, f0, a phenome duration, and a phoneme sequence that is processed by a signal generation unit of the back-end subsystem. The signal generation unit includes the neural network interacting with a pre-existing knowledgebase of phenomes to generate audible speech from the input vectors. The technique applies an error signal from the neural network to correct imperfections of the pre-existing knowledgebase of phenomes to generate audible speech signals. Speech signal specific modelling techniques in combination with applied psychoacoustic principles drive training efficiency of neural networks with positive impact on quality of generated speech signals.

Abstract: A soundalike system to improve speech synthesis by training a text to speech engine on a voice like the target speakers voice.

Abstract: Disclosed are a system for grasping keyword extraction based speech content on recorded voice data, an indexing method using the system, and a method for grasping speech content. An indexing unit receives voice data, performs per-frame voice recognition with reference to a phoneme to form a phoneme lattice, generates divided indexing information for a frame of a limited time configured with a plurality of frames, and stores the same in an indexing database, the divided indexing information including a phoneme lattice formed for each frame of the limited time.

Abstract: According to one embodiment, a transcription support system supports transcription work to convert voice data to text. The system includes a first storage unit configured to store therein the voice data; a playback unit configured to play back the voice data; a second storage unit configured to store therein voice indices, each of which associates a character string obtained from a voice recognition process with voice positional information, for which the voice positional information is indicative of a temporal position in the voice data and corresponds to the character string; a text creating unit that creates the text in response to an operation input of a user; and an estimation unit configured to estimate already-transcribed voice positional information indicative of a position at which the creation of the text is completed in the voice data based on the voice indices.

Abstract: The systems, devices, and processes described herein may asynchronously transfer audio signals from a voice-controlled device to a remote device. The audio signals may correspond to sound that is captured by multiple microphones of the voice-controlled device, which may then process the audio signals. The audio signals may also be transferred to the remote device for processing. Moreover, a determination of whether the voice-controlled device or the remote device is to process the audio signals may be based at least in part on the bandwidth of a network communicatively coupled to the voice-controlled device. The voice-controlled device may also cache and log the audio signals, and then asynchronously stream the audio signals to the remote device after the audio signals are initially processed, which may be based on the bandwidth of the network. The remote device may utilize the unprocessed audio signals to improve subsequent processing of audio signals.

Abstract: The present invention discloses a voiceprint verification method, apparatus, storage medium and device.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating representations of input sequences. One of the methods includes receiving a grapheme sequence, the grapheme sequence comprising a plurality of graphemes arranged according to an input order; processing the sequence of graphemes using a long short-term memory (LSTM) neural network to generate an initial phoneme sequence from the grapheme sequence, the initial phoneme sequence comprising a plurality of phonemes arranged according to an output order; and generating a phoneme representation of the grapheme sequence from the initial phoneme sequence generated by the LSTM neural network, wherein generating the phoneme representation comprises removing, from the initial phoneme sequence, phonemes in one or more positions in the output order.

Abstract: Embodiments of methods, systems, and storage medium associated with providing user records associated with characteristics that may be used to identify the user are disclosed herein. In one instance, the method may include obtaining features of an individual, determining identifying characteristics associated with the obtained features, and initiating a search for a record associated with the individual based in part on the identifying characteristics associated with the obtained features, and, based on a result of the search, a verification of the record associated with the individual. The method may further include receiving at least a portion of the record associated with the individual, based at least in part on a result of the verification. The verification may be based in part on a ranking associated with the record. Other embodiments may be described and/or claimed.

Abstract: Disclosed are a system for grasping keyword extraction based speech content on recorded voice data, an indexing method using the system, and a method for grasping speech content. An indexing unit receives voice data, performs per-frame voice recognition with reference to a phoneme to form a phoneme lattice, generates divided indexing information for a frame of a limited time configured with a plurality of frames, and stores the same in an indexing database, the divided indexing information including a phoneme lattice formed for each frame of the limited time.

Abstract: Systems and processes for automatic accent detection are provided. In accordance with one example, a method includes, at an electronic device with one or more processors and memory, receiving a user input, determining a first similarity between a representation of the user input and a first acoustic model of a plurality of acoustic models, and determining a second similarity between the representation of the user input and a second acoustic model of the plurality of acoustic models. The method further includes determining whether the first similarity is greater than the second similarity. In accordance with a determination that the first similarity is greater than the second similarity, the first acoustic model may be selected; and in accordance with a determination that the first similarity is not greater than the second similarity, the second acoustic model may be selected.

Abstract: A first set of characters may be received in response to a user input for text prediction. An estimate may be generated indicating what second set of characters will be inputted. The generating an estimate may be based on at least receiving data from a second user device. At least some of the data may not be located within the second user device's text dictionary. At least some of the data may be provided to the first user device.

Abstract: A system and method for syntactic re-ranking of possible transcriptions generated by automatic speech recognition are disclosed. A computer system accesses acoustic data for a recorded spoken language and generates a plurality of potential transcriptions for the acoustic data. The computer system scores the plurality of potential transcriptions to create an initial likelihood score for the plurality of potential transcriptions. For a particular potential transcription in the plurality of transcriptions, the computer system generates a syntactical likelihood score. The computer system creates an adjusted score for the particular potential transcription by combining the initial likelihood score and the syntactic likelihood score for the particular potential transcription.

Abstract: System and methods for controlling a user experience are described. In an aspect, an interface can comprise an interface device for rendering content to a user, a sensor having a gesture zone associated therewith configured to detect a dexterous gesture of a user within the gesture zone and generate a sensor signal representing the dexterous gesture. A processor may be provided in communication with the sensor and the interface device. The processor receives the sensor signal, analyzes the sensor signal to determine a control action associated with the detected dexterous gesture of the user, and configures the user interface based upon the determined control action of the user.

Abstract: An input signal may be classified by computing correlations between feature vectors of the input signal and feature vectors of reference signals, wherein the reference signals correspond to a class. The feature vectors of the input signal and/or the reference signals may be segmented to identify portions of the signals before performing the correlations. Multiple correlations of the segments may be combined to produce a segment score corresponding to a segment. The signal may then be classified using multiple segment scores, for example by comparing a combination of the segment scores to a threshold.

Abstract: A first set of characters may be received in response to a user input for text prediction. An estimate may be generated indicating what second set of characters will be inputted. The generating an estimate may be based on at least receiving data from a second user device. At least some of the data may not be located within the second user device's text dictionary. At least some of the data may be provided to the first user device.

Abstract: A first set of characters may be received in response to a user input for text prediction. An estimate may be generated indicating what second set of characters will be inputted. The generating an estimate may be based on at least receiving data from a second user device. At least some of the data may not be located within the second user device's text dictionary. At least some of the data may be provided to the first user device.

Abstract: A method of providing a Graphical User Interface (GUI) is provided. The method includes receiving an image to be used for the GUI, and selecting one or more object areas to be used for the GUI from the image, setting a GUI icon corresponding to the selected object area, and outputting the set GUI as a home screen for executing at least one of application on a display of the terminal.

Abstract: The invention concerns a method for generating a signature of a musical audio signal of a given duration, the method comprising the following steps: —modelling (104) the musical audio signal to obtain, for each frequency band of a set of n frequency bands, a diagram representing the energy of the audio signal for the frequency band, on the basis of the time during said given duration; —determining (103) musical transition times tk of the audio signal during the given duration; —associating (105) each musical transition time tk with an item of local information comprising a vector of n values representative, respectively, of the energy of the audio signal in each of the n diagrams obtained between musical transition time tk and a subsequent musical transition time tk+1 and/or a vector of n values representative, respectively, of the energy of the audio signal in each of the n diagrams obtained between musical transition time tk and a preceding musical transition time tk?1; —determining (106), on the basis of t

Abstract: In one embodiment, a method includes generating a waveform fingerprint based on captured ambient audio data; calculating a location of the computing system; sending the generated waveform fingerprint and the location to a server; and receiving instructions from the server to adjust one or more device settings of an output device of the computing system. The instructions being based at least in part on identifying one or more audio fingerprints that match the generated waveform fingerprint and correlating one or more of the identified audio fingerprints to a physical environment of the computing system. The method also includes adjusting one or more of the device settings of the output device of the computing system in accordance with the received instructions.

Abstract: In one embodiment synthetic training data items are generated, each comprising a) a textual representation of a synthetic sentence and b) one or more transcodes of the synthetic sentence comprising one or more actions and one or more entities associated with the one or more actions. For each synthetic training data item, the textual representation of the synthetic sentence is converted into a sequence of phonemes that represent the synthetic sentence. A first machine learning model is then trained as a transcoder that determines transcodes comprising actions and associated entities from sequences of phonemes, wherein the training is performed using a first training dataset comprising the plurality of synthetic training data items that comprise a) sequences phonemes that represent synthetic sentences and b) transcodes of the synthetic sentences. The transcoder may be used in a conversational agent.

Abstract: Computing devices and methods for providing speech recognition data from one computing device to another device are disclosed. In one disclosed embodiment, audio input is received at a client device and processed to generate speech recognition data. An estimated confidence level is determined for a portion of the data, where the confidence level exceeds a predetermined confidence threshold corresponding to a valid result. At least one statistically improbable characteristic associated with the portion of data is identified. Based on identifying the statistically improbable characteristic, the portion of data is provided to a server computing device for evaluation.

Abstract: A method for evaluating strength of an audio password by an electronic device is described. The method includes obtaining an audio signal captured by one or more microphones. The audio signal includes an audio password. The method also includes evaluating the strength of the audio password based on measuring one or more unique characteristics of the audio signal. The method further includes informing a user that the audio password is weak based on the evaluation of the strength of the audio password.

Abstract: In an augmented reality and/or a virtual reality system, virtual annotation surfaces, or virtual sheets, or virtual whiteboards, may be materialized in response to a detected gesture. A user may annotate, adjust, store, review and revise the virtual annotation surfaces, and allow for collaboration with other users, while in the current virtual environment, and/or within another virtual environment, and/or outside of the virtual environment.

Abstract: Altering properties of rendered objects and/or mixed reality environments utilizing control points associated with the rendered objects and/or mixed reality environments is described. Techniques described can include detecting a gesture performed by or in association with a control object. Based at least in part on detecting the gesture, techniques described can identify a target control point that is associated with a rendered object and/or a mixed reality environment. As the control object moves within the mixed reality environment, the target control point can track the movement of the control object. Based at least in part on the movement of the control object, a property of the rendered object and/or the mixed reality environment can be altered. A rendering of the rendered object and/or the mixed reality environment can be modified to reflect any alterations to the property.

Abstract: An audio file analyzer computing system includes technologies to, among other things, localize audio events of interest (such as speakers of interest) within an audio file that includes multiple different classes (e.g., different speakers) of audio. The illustrative audio file analyzer computing system uses a seed segment to perform a semi-supervised diarization of the audio file. The seed segment is pre-selected, such as by a human person using an interactive graphical user interface.

Abstract: Efficient empirical determination, computation, and use of an acoustic confusability measure comprises: (1) an empirically derived acoustic confusability measure, comprising a means for determining the acoustic confusability between any two textual phrases in a given language, where the measure of acoustic confusability is empirically derived from examples of the application of a specific speech recognition technology, where the procedure does not require access to the internal computational models of the speech recognition technology, and does not depend upon any particular internal structure or modeling technique, and where the procedure is based upon iterative improvement from an initial estimate; (2) techniques for efficient computation of empirically derived acoustic confusability measure, comprising means for efficient application of an acoustic confusability score, allowing practical application to very large-scale problems; and (3) a method for using acoustic confusability measures to make principled

Abstract: A method and system are provided for heterogeneous log analysis. The method includes performing hierarchical log clustering on heterogeneous logs to generate a log cluster hierarchy for the heterogeneous logs. The method further includes performing, by a log pattern recognizer device having a processor, log pattern recognition on the log cluster hierarchy to generate log pattern representations. The method also includes performing log field analysis on the log pattern representations to generate log field statistics. The method additionally includes performing log indexing on the log pattern representations to generate log indexes.

Abstract: A computing device is described that outputs, for display, a graphical keyboard comprising a plurality of keys. The computing device receives, an indication of a gesture detected at a presence-sensitive input device. The computing device determines, based at least in part on the indication of the gesture and at least one characteristic of the gesture, one or more keys from the plurality of keys. The computing device determines a character string based on the one or more keys from the plurality of keys. In response to determining that the character string is not included in a lexicon and a spatial model probability associated with the one or more keys from the plurality of keys exceeds a probability threshold, the computing device outputs, for display, the character string.

Abstract: Techniques are provided for reverberation compensation for far-field speaker recognition. A methodology implementing the techniques according to an embodiment includes receiving an authentication audio signal associated with speech of a user and extracting features from the authentication audio signal. The method also includes scoring results of application of one or more speaker models to the extracted features. Each of the speaker models is trained based on a training audio signal processed by a reverberation simulator to simulate selected far-field environmental effects to be associated with that speaker model. The method further includes selecting one of the speaker models, based on the score, and mapping the selected speaker model to a known speaker identification or label that is associated with the user.

Abstract: Technologies are described herein for enhancing a user presence status determination. Visual data may be received from a depth camera configured to be arranged within a three-dimensional space. A current user presence status of a user in the three-dimensional space may be determined based on the visual data. A previous user presence status of the user may be transformed to the current user presence status, responsive to determining the current user presence status of the user.

Abstract: Providing Internet application services includes acquiring an invocation command for an Internet application service and opening a service page corresponding to the Internet application service, generating an Internet application service invocation result based on an Internet application service invocation operation performed by a user on the service page, sending the Internet application service invocation result to the Internet application service provider and causing the Internet application service provider to generate invocation result card model data based on the Internet application service invocation result, generating an invocation result display card, and displaying the one or more other Internet application services to the user.

Abstract: The disclosed embodiments illustrate method and system of processing multimedia content to generate a text transcript. The method includes segmenting each of a set of text frames to determine spatial regions. The method further includes extracting one or more keywords from each of the determined spatial regions. The method further includes determining the first set of keywords from the extracted one or more keywords based on filtering of one or more off-topic keywords from the extracted one or more keywords. The method further includes extracting a second set of keywords based on the determined first set of keywords. The method further includes generating a graph between each of a first set of keywords and one or more of a second set of keywords. The method further includes dynamically generating the text transcript of audio content in the multimedia content based on the generated graph.

Abstract: Location-based conversational understanding may be provided. Upon receiving a query from a user, an environmental context associated with the query may be generated. The query may be interpreted according to the environmental context. The interpreted query may be executed and at least one result associated with the query may be provided to the user.

Abstract: Systems and methods are described for personifying communications. According to at least one embodiment, the computer-implemented method for personifying a natural-language communication includes observing a linguistic pattern of a user. The method may also include analyzing the linguistic pattern of the user and adapting the natural-language communication based at least in part on the analyzed linguistic pattern of the user. In some embodiments, observing the linguistic pattern of the user may include receiving data indicative of the linguistic pattern of the user. The data may be one of verbal data or written data. Written data may include at least one of a text message, email, social media post, or computer-readable note. Verbal data may include at least one of a recorded telephone conversation, voice command, or voice message.

Abstract: Provided are a method and an apparatus for performing exact start and end recognition of voice based on video recognition. The method includes determining whether a speech starts based on at least one of first video and audio data before conversion into a voice recognition mode, converting into the voice recognition mode and generating second audio data including a voice command, when it is determined that speech starts, and determining whether the speech is terminated based on at least one of second video and audio data after conversion into the voice recognition mode.

Abstract: The present invention includes systems and methods for sending social media messages without the need for keyboard inputs. A microphone captures live audio speech data and transmits the audio data to a processing unit. The processing unit converts the audio to speech data. The processing unit also removes censored words, emphasizes key words, and edits that data to include product and promotional messages where appropriate. The processing unit then uses code words contained in the speech data to send the speech data to the appropriate social media outlets for output.