Abstract: Disclosed embodiments relate to extracting classification information from input text.

Abstract: A method optimizes questions to retain engagement. The method includes generating, using a machine learning model, a churn risk from user interaction data. The method includes selecting, when the churn risk satisfies a threshold, a field, from multiple fields, using multiple prediction confidences corresponding to multiple prediction values generated for the multiple fields. The method includes obtaining a prediction value for the field and obtaining a question, corresponding to the field, using the prediction value. The method includes presenting the question and receiving a user input in response to the question.

Abstract: Implementations described herein include detecting a stream of audio data that captures a spoken utterance of the user and that captures ambient noise occurring within a threshold time period of the spoken utterance being spoken by the user. Implementations further include processing a portion of the audio data that includes the ambient noise to determine ambient noise classification(s), processing a portion of the audio data that includes the spoken utterance to generate a transcription, processing both the transcription and the ambient noise classification(s) with a machine learning model to generate a user intent and parameter(s) for the user intent, and performing one or more automated assistant actions based on the user intent and using the parameter(s).

Abstract: The present disclosure provides a summary generation model training method and apparatus, a device and a storage medium, and relates to the field of computer technologies, and in particular, to the field of artificial intelligence such as natural language processing and deep learning. The summary generation model training method includes: acquiring a document representation corresponding to a document sample; constructing, based on the document representation, a summary representation corresponding to the document representation, the summary representation including a positive summary representation and a negative summary representation; and constructing a total contrastive loss function based on the document representation, the positive summary representation and the negative summary representation, and training a summary generation model based on the total contrastive loss function. The present disclosure may improve accuracy of the summary generation model.

Abstract: An embodiment for creating a dialog based on anticipated questions for database driven conversations is provided. The embodiment may include receiving content from a database. The embodiment may also include identifying one or more schemas, one or more entities, and relational data in the content. The embodiment may further include identifying a semantic type and a number of distinct entries for each entity. The embodiment may also include presenting choices for one or more query targets and one or more filtering conditions to a user. The embodiment may further include prompting the user for one or more annotations and one or more clarifying questions for each chosen query target and filtering condition. The embodiment may also include generating a plurality of modular phrases and combining the plurality of modular phrases into one or more sentences and paraphrases of the one or more sentences.

Abstract: A method may including obtaining a voice transcript corpus and a chat transcript corpus, extracting voice transcript sentences from the voice transcript corpus and chat transcript sentences from the chat transcript corpus, encoding, by a series of neural network layers, the voice transcript sentences to generate voice sentence vectors, encoding, by the series of neural network layers, the chat transcript sentences to generate chat sentence vectors, determining, for each voice sentence vector, a matching chat sentence vector to obtain matching voice-chat vector pairs, and adding, to a parallel corpus, matching voice-chat sentence pairs using the matching voice-chat vector pairs. Each of the matching voice-chat sentence pairs may include a voice transcript sentence and a matching chat transcript sentence. The method may further include training a disfluency remover model using the parallel corpus.

Abstract: A data processing method based on simultaneous interpretation, applied to a server in a simultaneous interpretation system, including: obtaining audio transmitted by a simultaneous interpretation device; processing the audio by using a simultaneous interpretation model to obtain an initial text; transmitting the initial text to a user terminal; receiving a modified text fed back by the user terminal, the modified text being obtained after the user terminal modifies the initial text; and updating the simultaneous interpretation model according to the initial text and the modified text.

Abstract: Described herein are systems and methods for facilitating the information entry and task updates to a task database in a cloud server. The task database is in synchronization with a customer relationship management (CRM) system. The systems and methods described herein enable users to update the task database and enter information into the task database in a timely manner such that the task database can stay updated. The updated database can be used to construct a suggested task set at the beginning of a period of time to meet a preset target sales value for the end of the period of time. In one embodiment, a system includes a mobile application to capture contact information from a physical medium as an image, and to send the image to a cloud server, where a trained neural network model is to extract contact details and send the contact details back to the mobile application for editing and confirmation by a user.

Abstract: A system and method for concurrent multi-path processing of audio signals for automatic speech recognition is presented. Audio information defining a set of audio signals may be obtained (502). The audio signals may convey mixed audio content produced by multiple audio sources. A set of source-specific audio signals may be determined by demixing the mixed audio content produced by the multiple audio sources. Determining the set of source-specific audio signals may comprises providing the set of audio signals to both a first signal processing path and a second signal processing path (504). The first signal processing path may determine a value of a demixing parameter for demixing the mixed audio content (506). The second signal processing path may apply the value of the demixing parameter to the individual audio signals of the set of audio signals (508) to generate the individual source-specific audio signals (510).

Abstract: A control unit is adapted to perform a process related to an electric power tool, based on sound input to a microphone. The control unit performs a process related to the electric power tool, based on a result of subjecting input sound to a sound recognition process or a result of subjecting the sound to sound analysis. When a user operation switch is in an on state, the control unit does not perform a sound recognition process, and, when the user operation switch is in an off state, the control unit performs a sound recognition process. Further, when the electric power tool is not in a state of being gripped by the user, the control unit does not perform a sound recognition process, and, when the electric power tool is in a state of being gripped by the user, the control unit performs a sound recognition process.

Abstract: Systems and techniques are provided for sound model localization within an environment. Sound recordings of sounds in the environment may be received from devices in the environment. Preliminary labels for the sound recordings may be determined using pre-trained sound models. The preliminary labels may have associated probabilities. Sound clips with preliminary labels may be generated based on sound recordings that have preliminary labels whose probability is over a high-recall threshold for the pre-trained sound model that determined the preliminary label. The sound clips with preliminary labels may be sent to a user device. Labeled sound clips may be received from the user device. The labeled sound clips may be based on the sound clips with preliminary labels. Training data sets may be generated for the pre-trained sound models using the labeled sound clips. The pre-trained sound models may be trained using the training data sets to generate localized sound models.

Abstract: Systems, apparatuses, methods, and computer program products are disclosed for determining robustness information for an NLP model. Modification rules, such as replacement rules and/or insertion rules, are used to generate instances of modified test data based on instances of test data that comprise words and have a syntax and a semantic meaning. The instances of test data and modified test data are provided to the NLP model and the output of the NLP model is analyzed to determine output changing instances of modified test data, which are instances of modified test data yielded output from the NLP model that is different and/or not similar to the output yielded from the NLP model for the corresponding instance of test data. Robustness information for the NLP model is determined based at least in part on the output changing instances of modified test data.

Abstract: In general, techniques are described for psychoacoustic audio coding of ambisonic audio data. A device comprising a memory and one or more processors may be configured to perform the techniques. The memory may store the bitstream that includes an encoded audio object and a corresponding spatial component that defines spatial characteristics of the encoded foreground audio signal. The encoded foreground audio signal may include a coded gain and a coded shape. The one or more processors may perform a gain and shape synthesis with respect to the coded gain and the coded shape to obtain a foreground audio signal, and reconstruct, based on the foreground audio signal and the spatial component, the ambisonic audio data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a generative neural network to convert conditioning text inputs to audio outputs using energy scores.

Abstract: Methods, apparatus, systems, and computer-readable media are provided for using shortcut command phrases to operate an automated assistant. A user of the automated assistant can request that a shortcut command phrase be established for causing the automated assistant to perform a variety of different actions. In this way, the user does not necessarily have to provide an individual command for each action to be performed but, rather, can use a shortcut command phrase to cause the automated assistant to perform the actions. The shortcut command phrases can be used to control peripheral devices, IoT devices, applications, websites, and/or any other apparatuses or processes capable of being controlled through an automated assistant.

Abstract: A system, method, and computer-readable medium for performing a data center monitoring and management operation. The data center monitoring and management operation includes: selecting a reference phrase; presenting the reference phrase to a user; generating a voice signature the reference phrase when the reference phrase is vocalized by the user; storing the voice signature for reference phrase within a data center monitoring and management console; instructing the user to recite a subset of words from the reference phrase; and, granting access to the data center monitoring and management console when the subset of words match respective voice signatures stored within the data center monitoring and management console.

Abstract: A method for facilitating a multiparty conversation is disclosed. An electronic device using the method may facilitate a multiparty conversation by identifying participants of a conversation, localizing relative positions of the participants, detecting speeches of the conversation, matching one of the participants to each of the detected speeches according to the relative positions of the participants, counting participations of the matched participant in the conversation, identifying a passive subject from all the participants according to the participations of all the participants in the conversation, finding a topic of the conversation between the participants, and engaging the passive subject by addressing the passive subject and speaking a sentence related to the topic.

Abstract: An acoustic event detection system may employ self-supervised federated learning to update encoder and/or classifier machine learning models. In an example operation, an encoder may be pre-trained to extract audio feature data from an audio signal. A decoder may be pre-trained to predict a subsequent portion of audio data (e.g., a subsequent frame of audio data represented by log filterbank energies). The encoder and decoder may be trained using self-supervised learning to improve the decoder's predictions and, by extension, the quality of the audio feature data generated by the encoder. The system may apply federated learning to share encoder updates across user devices. The system may fine-tune the classifier to improve inferences based on the improved audio feature data. The system may distribute classifier updates to the user device(s) to update the on-device classifier.

Abstract: An audio processing method by which input data are obtained that includes first sound data representing first components of a first frequency band, included in a first sound corresponding to a first sound source, second sound data representing second components of the first frequency band, included in a second sound corresponding to a second sound source, and mix sound data representing mix components of an input frequency band including a second frequency band, the mix components being included in a mix sound of the first sound and the second sound. The input data are then input to a trained estimation model, to generate at least one of first output data representing first estimated components within an output frequency band including the second frequency band, included in the first sound, or second output data representing second estimated components within the output frequency band, included in the second sound.

Abstract: A method for semantic recognition includes: in response to performing semantic analysis on information acquired by a terminal, a sentence to be processed is acquired; word recognition is performed on the sentence, to obtain a plurality of words and part-of-speech information corresponding to each of the words; a target set update operation is determined with a word processing model, according to one or more words to be input, part-of-speech information of the words to be input, and a dependency relationship of a first word. The word processing model is configured to calculate first and second feature vectors according to a word feature vector of the words to be input, a part-of-speech feature vector of the part-of-speech information and a relationship feature vector of the dependency relationship of the first word, calculate confidence levels of the preset set update operations according to the first feature vector and the second feature vector.