Abstract: A method and system for modifying an audible command is provided. The method includes continuously receiving audible commands associated with a context of interactions between a user and individuals. The audible commands are analyzed with respect to associated actions and user attributes of the audible commands are identified. Specified information required for executing each command of the audible commands and portions of the specified information associated with specified individuals of the individuals are determined. Digital audio samples of the user are retrieved and assigned to the portions of the specified information with respect to each command. The associated actions are modified with respect to the specified individuals and self-learning software code comprising the modified actions is generated and executed such that the commands are executed with respect to the modified actions.

Abstract: Devices and techniques are generally described for rejecting false invocations of speech processing skills. In various examples, utterance data comprising automatic speech recognition (ASR) data and natural language understanding (NLU) data may be received. In some examples, ASR confidence data indicating a confidence level of the ASR data may be received. In further examples, NLU confidence data indicating a confidence level of the NLU data may be received. A machine learning model may determine, based at least in part on the ASR confidence data and the NLU confidence data, first false invocation data indicating a likelihood of false invocation of a speech processing skill. In some examples, a first directive may be sent to the speech processing system based at least in part on the first false invocation data. The first directive may be effective to cause the speech processing system to end a current dialog session.

Abstract: Disclosed is a semantic sentiment analysis method fusing in-depth features and time sequence models, including: converting a text into a uniformly formatted matrix of word vectors; extracting local semantic emotional text features and contextual semantic emotional text features from the matrix of word vectors; weighting the local semantic emotional text features and the contextual semantic emotional text features by using an attention mechanism to generate fused semantic emotional text features; connecting the local semantic emotional text features, the contextual semantic emotional text features and the fused semantic emotional text features to generate global semantic emotional text features; and performing final text emotional semantic analysis and recognition by using a softmax classifier and taking the global semantic emotional text features as input.

Abstract: The present disclosure relates to a method and system for robust and efficient language identification. The method includes: receiving the speech signal; partitioning the speech signal into a plurality of audio frames; extracting features of the plurality of audio frames; determining, using a neural network, a variable associated with the language identity and one or more auxiliary attributes of the speech signal, for each of the plurality of audio frames; determining scores of the plurality of audio frames based on the extracted features; and determining the language identity of the speech signal based on the variables and scores determined for the plurality of audio frames.

Abstract: The disclosed embodiments provide a system for performing a natural language search. During operation, the system applies a first machine learning model to a natural language query to predict one or more search intentions associated with the natural language query. Next, the system applies a second machine learning model to the natural language query to produce one or more search parameters associated with a first intention in the search intention(s), wherein the search parameter(s) include a field and a value of the field. The system then performs a first search of a first vertical associated with the first intention using the search parameter(s). Finally, the system generates a ranking containing a first set of search results from the first search of the first vertical and outputs at least a portion of the ranking in a response to the natural language query.

Abstract: A question answering device includes: a general word vector converter converting a question and an answer to semantic vectors in accordance with general context; a general sentence level CNN 214, in response to similarities of semantic vectors between words in question and answer and to strength of causality between the words, for weighting each semantic vector to calculate sentence level representations of the question and the answer; a general passage level CNN 218, in response to similarity between sentence level representations of question and answer, and to strength of relation of vectors in the sentence level representations viewed from causality, for weighting the sentence level representation to calculate a passage level representation for the question and answer passage; and a classifier determining whether or not an answer is a correct answer, based on the similarities between outputs from CNNs 214 and 218.

Abstract: A computer-implemented method includes identifying a theme for a poetic composition. The method also includes building a datastore of text taxonomies that provide associative differences between two words in a plurality of words. The method also includes analyzing the associative differences to determine a distance from a nexus for each of the plurality of words, wherein the nexus is determined according to the theme, and wherein the distance from the nexus to one of the plurality of words describes a suitability of the one of the plurality of words for use in the poetic composition for the theme. The method also includes determining at least one associative word according to the distance from the nexus. The method also includes determining a rhyming word from the plurality of words according to the at least one associative word.

Abstract: A computer-executed conversational system utilizing adaptive dialogs to provide dynamic modification of a conversation structure. The adaptive dialogs facilitate an adaptive and dynamic way to model conversations. A conversational rule may be associated with a trigger received at an adaptive dialog. The conversational rule may populate a plan with steps to provide the conversational structure of the conversational system. Modifying triggers may be received that allow conversational rules to amend the plan including adding steps, removing steps, or modifying steps from the plan. The adaptive dialogs and/or components thereof may be provided as constructible elements in a platform for building a bot such as a developer Software Development Kit (SDK) or the like. In turn, sophisticated conversational modeling may be conveniently facilitated to provide features such as a dialog dispatcher, the ability to handle interruptions, and to build a pluggable, extensible conversational system.

Abstract: In a system and method for audio analysis in a cloud-based computerized an authentication (RTA) manager micro-service may send an audio packet to a voice processor micro-service. The voice processor may extract features of the audio. The RTA manager may obtain the extracted features from the voice processor; calculate, based on the extracted features, a quality grade of the audio packet, and send the extracted features to an at least one voice biometrics engine if the quality grade is above a threshold. Each of the at least one voice biometrics engines may be configured to generate a voiceprint of the audio packet, based on the extracted features of the audio packet and to perform at least one of: authenticate a speaker, detect fraudsters, and enrich a previously stored voiceprint of the speaker with the voiceprint of the audio packet.

Abstract: In various example embodiments, a system and method for correction of user input are presented. In one embodiment, a method includes receiving a plurality of user strings, selecting one or more string pairs from the plurality of user strings based on a character operator difference between the first string and the second string being below a threshold number, filtering the one or more string pairs to generate a filtered set of strings pairs representing corrections, and correcting user input in a different session by replacing input that matches a first string in a filtered string pair with a second string in the filtered string pair.

Abstract: Conversational event modeling for determining how to process an input in a dialog stack of a computer-executed conversational system. Receipt of an input at an active dialog in a dialog stack may result in generation of an event corresponding to the input. The event may be propagated through the dialog stack. Dialogs in the dialog stack may generate a confidence value in relation to processing the input, and selection of the dialog in the dialog stack for use in processing the input may be based at least in part on the confidence values. In turn, the conversational system may facilitate improved handling of ambiguous or unrelated inputs at dialogs by propagating the event associated with such an input through the dialog stack. The event creation and processing functions may be provided as parameters for dialogs in a modular dialog creation system.

Abstract: A computing system and related techniques for selecting content to be automatically converted to speech and provided as an audio signal are provided. A text-to-speech request associated with a first document can be received that includes data associated with a playback position of a selector associated with a text-to-speech interface overlaid on the first document. First content associated with the first document can be determined based at least in part on the playback position, the first content including content that is displayed in the user interface at the playback position. The first document can be analyzed to identify one or more structural features associated with the first content. Speech data can be generated based on the first content and the one or more structural features.

Abstract: Embodiments of the present disclosure relate to a method and apparatus for outputting information. The method includes: outputting a to-be-read audio in response to receiving a reading instruction from a user; acquiring an actually read audio obtained by reading the to-be-read audio by the user; performing speech recognition on the actually read audio to obtain a recognition result; calculating a similarity between the actually read audio and the to-be-read audio based on a character string corresponding to the recognition result and a character string corresponding to the to-be-read audio; determining, from a predetermined set of similarity intervals, a similarity interval to which the calculated similarity belongs; and outputting a reading evaluation corresponding to the determined similarity interval. The embodiment may help a reader to improve the learning efficiency and learning interest, thereby improving the rate of a user using a device.

Abstract: A system for processing user utterances and/or text based queries that tracks context data of a current dialog between the system and the user, can determine values for slots by determining if the value was mentioned in a previous turn of the dialog. The system may employ a question-answer model and construct a question including the domain name, slot name, and potential slot values. The system may process the previous turns of the dialog and the constructed question to determine the slot value. The system may track a dialog that relates to multiple domains and may determine the slot value based on a slot value that is mentioned in relation to another domain.

Abstract: Certain aspects of the present disclosure provide techniques for detecting personally identifiable information, including: receiving a plurality of text strings, each text string of the plurality of text strings associated with a user support session; providing the plurality of text strings to one or more bidirectional long short-term memory (BiLSTM) neural network models; receiving output from the one or more BiLSTM neural network models, the output indicating one or more text data elements in the plurality of text strings comprising predicted personally identifiable information; redacting the one or more text data elements comprising the predicted personally identifiable information from the plurality of text strings to form redacted text strings; and providing, to a data repository, the redacted text strings.

Abstract: An apparatus that includes a microphone and a processor. The processor is configured to receive, via the microphone, audio comprising voice of a person, and determine whether the received audio is an actual voice or a synthesized voice. The apparatus also provides a first notification indicating that the received audio is the actual voice when the received audio is the actual voice, and provides a second notification indicating that the received audio is the synthesized voice when the received audio is the synthesized voice.

Abstract: Implementations relate to mitigating client device latency in rendering of remotely generated automated assistant content. Some of those implementations mitigate client device latency between rendering of multiple instances of output that are each based on content that is responsive to a corresponding automated assistant action of a multiple action request. For example, those implementations can reduce latency between rendering of first output that is based on first content responsive to a first automated assistant action of a multiple action request, and second output that is based on second content responsive to a second automated assistant action of the multiple action request.

Abstract: A phrase may be received that includes a plurality of tokens in a natural language format. A plurality of levels relating to dependencies between tokens of the plurality of tokens within the phrase is determined. A matrix structure is generated for the phrase. The matrix structure utilizes a plurality of rows and a plurality of columns to store data of the phrase. The plurality of rows and the plurality of columns each indicate one of an order of tokens of the plurality of tokens or levels of the plurality of levels.

Abstract: A self-learning translation-engine brokering system characterizes a set of language-translation engines by associating each engine with values of a set of engine parameters. The system receives a request to translate text or speech input, along with a set of weightings that identify the relative importance of each engine parameter to the translation requester. The system formats the input into a quantifiable engine-agnostic form and performs an optimization procedure that finds the best fit between the source's weightings and the sets of engine parameters. The system directs the input to the best-fitting translation engine, receives the translated output from the selected engine, directs the output to the translation requester, and then determines how well the translated output meets user expectations. This feedback used to update the best-fitting engine's parametric values and to train the system to make more accurate selections in the future.

Abstract: Methods and apparatus related to determining an effect on dissemination of information related to an event based on a dynamic confidence level associated with the event. For example, an event and an event confidence level of the event may be determined based on a message of a user. An effect on dissemination of information related to the event may be determined based on the confidence level. A new confidence level may be determined based on additional data associated with the event and the effect on dissemination of information may be adjusted based on the new confidence level. In some implementations, the additional data may be based on a new message that is related to the message, such as a reply to the message.