Abstract: Systems and methods are disclosed herein for building contextual transcripts. A computing system may receive a textual transcript of a meeting that contains a variety of statements made by various attendees of the meeting, select the first statement made during the meeting, and determine which meeting attendee made the statement. A machine learning model corresponding to the particular attendee that has been trained using previously received statements by the particular attendee may be used on the utterance to determine the tone of the utterance. That tone may be recorded within the transcript and this process may be repeated for each utterance to build a contextual transcript.

Abstract: A method for retrieving a multi-turn dialogue, a storage medium, and an electronic device are provided. The detailed implementation includes: obtaining present question information and context information of the present question information; obtaining a plurality of pieces of candidate reply information of the present question information; generating a context feature based on the context information, and generating a plurality of candidate reply features based on the plurality of pieces of candidate reply information; obtaining consistency information between the context feature and each of the plurality of candidate reply features; selecting a piece of candidate reply information from the plurality of pieces of candidate reply information for replying based on the consistency information between the context feature and each of the plurality of candidate reply features.

Abstract: The present technology relates to an information processing apparatus, an information processing method, and a program which can curb occurrence of howling at the time of outputting vibration in response to an input sound. The information processing apparatus of one aspect of the present technology is an apparatus that generates, at the time of outputting vibration in response to an input sound from the outside, a vibration signal representing the vibration having a frequency different from a frequency of the input sound. The present technology can be applied to, for example, smartphones, smart watches, wearable apparatuses, cushions, and music experience apparatuses that vibrate in response to input sounds.

Abstract: A speaker verification method includes receiving audio data corresponding to an utterance, processing a first portion of the audio data that characterizes a predetermined hotword to generate a text-dependent evaluation vector, and generating one or more text-dependent confidence scores. When one of the text-dependent confidence scores satisfies a threshold, the operations include identifying a speaker of the utterance as a respective enrolled user associated with the text-dependent confidence score that satisfies the threshold and initiating performance of an action without performing speaker verification. When none of the text-dependent confidence scores satisfy the threshold, the operations include processing a second portion of the audio data that characterizes a query to generate a text-independent evaluation vector, generating one or more text-independent confidence scores, and determining whether the identity of the speaker of the utterance includes any of the enrolled users.

Abstract: An audio encoder for encoding an audio signal has: a first encoding processor for encoding a first audio signal portion in a frequency domain, having: a time frequency converter for converting the first audio signal portion into a frequency domain representation; an analyzer for analyzing the frequency domain representation to determine first spectral portions to be encoded with a first spectral resolution and second regions to be encoded with a second resolution; and a spectral encoder for encoding the first spectral portions with the first spectral resolution and encoding the second portions with the second resolution; a second encoding processor for encoding a second different audio signal portion in the time domain; a controller for analyzing and determining, which portion of the audio signal is the first audio signal portion encoded in the frequency domain and which portion is the second audio signal portion encoded in the time domain; and an encoded signal former for forming an encoded audio signal havi

Abstract: Embodiments provide a system and method for statistical subject identification. The system takes texts, videos, audios, and images as input for which subject needs to be identified. The system pre-process input data and generates n-grams and pre-processed text strings by removing stopwords, punctuations, selective POS tags and lemmatization. Frequency distribution of n-grams are computed, and weightage of n-grams is assigned. For each n-gram, sum of weights across all text strings is computed and a maximum weightage is identified. The computed value as a result of taking a ratio of two, is assigned to each of the n-grams. Values computed for the n-grams have a non-normal distribution, when observed statistically. Thus, the n-gram values are transformed to confidence value following a normal distribution. The system maps the n-gram domains using a domain lexicon. Finally, these domains are aggregated and converged for subject identification based on a pre-annotated mapping dictionary.

Abstract: The disclosure relates to system and method for processing user input using Natural Language Processing (NLP). The method includes generating, by an NLP model, a set of input intent maps associated with a user input. The method includes matching each of the set of input intent maps with each of a plurality of pre-stored sets of intent maps. Each of the plurality of pre-stored sets of intent maps is generated from a single predefined training input and is mapped to a predefined intent and a predetermined response. The method includes determining a distance of each of the set of input intent maps relative to each of the plurality of pre-stored sets of intent maps. Further, the method includes identifying a pre-stored intent map closest to the set of input intent maps and rendering the predetermined response mapped to the pre-stored sets of intent maps to the user.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer-readable media, for speech recognition using multi-dialect and multilingual models. In some implementations, audio data indicating audio characteristics of an utterance is received. Input features determined based on the audio data are provided to a speech recognition model that has been trained to output score indicating the likelihood of linguistic units for each of multiple different language or dialects. The speech recognition model can be one that has been trained using cluster adaptive training. Output that the speech recognition model generated in response to receiving the input features determined based on the audio data is received. A transcription of the utterance generated based on the output of the speech recognition model is provided.

Abstract: A method, including obtaining data indicative of respective perceived loudness levels for a plurality of hearing percepts respectively evoked at different current levels, and creating a map for the hearing prosthesis based on the obtained data by adjusting at least one of the respective current levels based on data of a respective perceived loudness for another current level.

Abstract: To be able to convert to a voice of the desired attribution.

Abstract: A method for execution by a computing device includes obtaining a phrase that includes string of words and generating a valid sequence of words utilizing the phrase. The method further includes identifying a set of identigens for each word of the valid sequence of words to produce sets of identigens. The method further includes identifying, for each identigen of the sets of identigens, a word type associated with phrase structure grammar rules to produce sets of identigen-type associations. The method further includes interpreting, utilizing the phrase structure grammar rules, the sets of identigen-type associations to produce an entigen group. The entigen group represents a most likely interpretation of the phrase.

Abstract: An electronic device and a method for controlling thereof is provided. The electronic device includes a memory storing a neural network model and a processor configured to input, to the neural network model, input data to obtain output data, and, based on comparison between first output data based on input first modality and second output data based on input second modality, in response to the second modality being input, the neural network model is trained to output the first modality corresponding to the first output data, and the second modality may include at least one masking element.

Abstract: A subgraph extraction means 71 extracts, from a document structure graph indicating a document structure, a subgraph as a part of the document structure graph on the basis of inter-word relationship information indicating a relationship between a word and a word. A rule creation means 72 creates a rule for extracting a subgraph having the same structure as the subgraph from the document structure graph. A knowledge addition means 73 extracts a subgraph from the document structure graph in accordance with the rule and adds the information indicated by the subgraph to the inter-word relationship information.

Abstract: Provided is a voice signal processing method. The method comprises the following steps: acquiring a status coefficient, a target voice signal to be processed, and a derivative signal of the target voice signal (S101); using the derivative signal to respectively calculate an initial echo loss corresponding to each signal frame in the target voice signal, then using the status coefficient to adjust the initial echo loss to acquire a target echo loss (S102); using the derivative signal to calculate a proximal-distal coherence value of the target voice signal (S103); judging whether the target echo loss matches with a single-talk echo state (S104); if so, recording the proximal-distal coherence value in the status statistical data, and updating statistical information (S105); and using the statistical information to update the status coefficient, and performing volume control on the target voice signal after the status coefficient is updated (S106).

Abstract: Disclosed is an apparatus and method for processing a multichannel audio signal. A multichannel audio signal processing method may include: generating an N-channel audio signal of N channels by down-mixing an M-channel audio signal of M channels; and generating a stereo audio signal by performing binaural rendering of the N-channel audio signal.

Abstract: Embodiments of the present invention provide methods, computer program products, and systems determine a user state based on received information. Embodiments of the present invention can, in response to determining a user state reaching a threshold value, perform an ameliorative action.

Abstract: Systems and methods disclosed herein include, while a microphone of a first networked microphone device is enabled, determining whether a first reference device is in a specific state, and in response to determining that the first reference device is in the specific state, disabling the microphone of the first networked microphone device. Some embodiments further include, while the microphone of the first networked microphone device is enabled, receiving a command to disable the microphone of the first networked microphone device via one of the microphone of the networked microphone device or a network interface of the networked microphone device, and in response to receiving the command to disable the microphone of the networked microphone device via one of the microphone of the networked microphone device or the network interface of the networked microphone device, disabling the microphone of the networked microphone device.

Abstract: A system may receive a communication from a user, which may include a vocal sample. The system may transform the vocal sample from a wavelength domain into a frequency domain. The system may determine a divergence of one or more amplitude values of the transformed frequency domain from a predetermined frequency distribution. According to some embodiments, the predetermined frequency distribution may be a Benford's distribution. When the divergence exceeds a predetermined threshold, the system may execute one or more security measures. The one or more security measures may include (i) transferring the user from an automated operator to a human operator, (ii) requiring second factor authentication from the user, and/or (iii) denying a user-initiated request.

Abstract: An approach is provided in which a method, system, and program product build a time series prediction model based on one or more relationships between a first set of keywords in a set of first news articles and a second set of keywords in a set of second news articles. The time series prediction model includes a time-based interest level adjustment corresponding to a publication time between the set of first news articles and the set second of news articles. The method, system, and program product use the time series prediction model to compute an inherited initial interest level of a third news article that includes a set of new keywords based on the set of new keywords and the time-based interest level adjustment. The method, system, and program product assign the inherited initial interest level to the third news article.

Abstract: The disclosed subject matter relates to a system and method for providing an automated assistant that retrieves information from a knowledge base in response to a user's natural language question. A user's natural language question voice is transformed into an intermediate representation. From the intermediate representation, a cypher query is generated which may be used to query the database. The query results are provided in response to the user. The transformation into the intermediate representation is database independent while the cypher query is dependent upon the database queried.