Abstract: Embodiments of the present invention provide computer implemented methods, computer program products and computer systems. For example, embodiments of the present invention can access one or more intents and associated entities from limited amount of speech to text training data in a single language. Embodiments of the present invention can locate speech to text training data in one or more other languages using the accessed one or more intents and associated entities to locate speech to text training data in the one or more other languages different than the single language. Embodiments of the present invention can then train a neural network based on the limited amount of speech to text training data in the single language and the located speech to text training data in the one or more other languages.

Abstract: An audio sample including speech and ambient sounds is transmitted to a vehicle computer. Recorded audio is received from the vehicle computer, the recorded audio including the audio sample broadcast by the vehicle computer and recorded by the vehicle computer and recognized speech from the recorded audio. The recognized speech and text of the speech are input to a machine learning program that outputs whether the recognized speech matches the text. When the output from the machine learning program indicates that the recognized speech does not match the text, the recognized speech and the text are included in a training dataset for the machine learning program.

Abstract: Disclosed herein are system, method, and computer program product embodiments for utilizing non-RAM memory to implement machine learning configured with a meta-learning training set (small dataset), to create a common-sense predictive language model, thus boosting the performance for downstream tasks. An embodiment operates by receiving a base sentence and perturbation sentences as an input and tokenizing the input to generate a sequence of tokens. Tokens of the semantic perturbation sentences are embedded with tokens of the base sentence as contextually similar tokens pairs to generate training data and classified to capture relationships of the base sentence and the perturbation sentences to generate a classification, which is used to train a language model.

Abstract: A voice call translation capability negotiation method and an electronic device are provided, and relate to the field of terminal technologies. The method includes: After a first electronic device establishes a communication link with a second electronic device, if a voice call translation function is enabled, the first electronic device receives first indication information sent by the second electronic device. The first indication information is used to indicate that a voice call translation function of the second electronic device is enabled. The first electronic device automatically disables the voice call translation function.

Abstract: In examples, a schema augmentation system for exploratory research leverages intelligence from a machine learning model to augment such tasks by leveraging intelligence derived from machine learning capabilities. Augmenting tasks include schematization of content, such as information units and groupings of information units. Based on the schematization of such content, semantic proximities for information units are determined. The semantic proximities may be used to identify and present potentially relevant information units, for example to accelerate the exploratory research task at hand. As such, users engaged in consumption of heterogeneous content (e.g., across client applications and/or content sources), may receive machine-augmented support to find potential information units.

Abstract: A customizable communication system and method of use are described for providing dialect and language options for users to employ during interactions between the user and a third-party application, thereby enhancing user experience. In some embodiments, the system allows a user to select a plurality of dialect and language preferences while interacting with a third-party application offering voice command technology. The selected dialect and language preference is used during the interaction between the user and the third-party application.

Abstract: Disclosed are an electronic device and a method of controlling the electronic device. An electronic device according to an embodiment may perform a method comprising: performing natural language understanding for a first text included in learning data, obtaining first information associated with a speech corresponding to the first text being uttered based on a result of the natural language understanding, obtain second information associated with an acoustic feature corresponding to the speech corresponding to the first text being uttered based on the first information, obtaining a plurality of speech signals corresponding to the first text by converting a first speech signal corresponding to the first text based on the first information and the second information, and training a speech recognition model based on the plurality of obtained speech signals and the first text.

Abstract: Methods and systems for training an intent classifier. For example, a question-intent tuple dataset comprising data samples is received. Each data sample has a question, an intent, and a task. A pre-trained language model is also received and fine-tuned by adjusting values of learnable parameters. Parameter adjustment is performed by generating a plurality of neural network models. Each neural network model is trained to predict at least one intent of the respective question having a same task value of the tasks of the question-intent tuple dataset. Each task represents a source of the question and the respective intent. The fine-tuned language model generates embeddings for training input data, the training input data comprising a plurality of data samples having questions and intents. Further, feature vectors for the data samples of the training input data are generated and used to train an intent classification model for predicting intents.

Abstract: Techniques for tuning parameters for machine learning models are described. Different values for a parameter are tested to determine the value that results in an optimized model. A parameter value may be selected for testing using a search algorithm based on how the model performs with respect to other values for the parameter. Different values may be tested until a stopping criterion (such as time for testing, number of trials, amount of enhancement in performance, etc.) is met. In some embodiments, the techniques may be used to determine parameter values for natural language processing models.

Abstract: The disclosure herein describes a system for interpreting text-based similarity between a seed item and a recommended item selected by a pre-trained language model from a plurality of candidate items based on semantic similarities between the seed item and the recommended item. The system analyzes similarity scores and contextual paragraph representations representing text-based descriptions of the seed item and recommended item to generate gradient maps and word scores representing the text-based descriptions. A model for interpreting text-based similarity utilizes the calculated gradients and word scores to match words from the seed item description with words in the recommended item description having similar semantic meaning. The word-pairs having the highest weight are identified by the system as the word-pairs having the greatest influence over the selection of the recommended item from the candidate items by the original pre-trained language model.

Abstract: In cases in which a confidence score of an inferred intent label is a predetermined threshold or less, an intent inference section searches for whether or not wording pertaining to a location, such as “on the door”, is present in a question. In cases in which a word relating to a location is present, the intent inference section consults individual function identification data associated with OM item codes in order to find intent labels including individual functions relevant to the location (such as “door”). In cases in which an intent label including an individual function relevant to the “door” is found, an OMA interaction control section consults QA data to find and acquire associated response information based on the found intent label and the OM item code, and notifies a HMI interaction control section of such response information.

Abstract: Described is a system and method that determines character sequences from speech, without determining the words of the speech, and processes the character sequences to determine sentiment data indicative of emotional state of a user that output the speech. The emotional state may then be presented or provided as an output to the user.

Abstract: The dialogue system includes a keyword acquisition unit configured to acquire an input key group containing one or a plurality of input keywords on the basis of an input of a character string, a category generation unit configured to classify the resulting sentence candidates into a plurality of categories on the basis of a comparison between the input key group acquired by the keyword acquisition unit and the storage key group contained in an FAQ database, an intra-category ranking determination unit configured to determine a priority ranking of the resulting sentence candidates within each of the categories, and a presentation unit configured to select a resulting sentence candidate of a highest priority ranking determined by the intra-category ranking determination unit from within a category of a highest priority ranking determined in advance, and present a response for prompting a user to make an additional input.

Abstract: Participants may use one or more devices for engaging in a meeting, such as phones, conferencing devices, and/or computers. The devices include microphones that capture speech for determining the presence of distinct participants. Speech signals originating from different participants, or microphones, may be determined and associated with the participants. For example, microphones may be directional and more sensitive to sound coming from one or more specific directions than sound coming from other directions. By associating an individual with a microphone, or set of microphones, overlapping voices may be disambiguated to provide clear voice streams that aid in producing a clear transcript indicating the speech of the participants, respectively. An identity of the participants may be determined using voiceprint and/or voice recognition techniques.

Abstract: A first named entity recognition (NER) system may be adapted to create a second NER system that is able to recognize a new named entity using few-shot learning. The second NER system may process support tokens that provide one or more examples of the new named entity and may process input tokens that may contain the new named entity. The second NER system may use a classifier of the first NER system to compute support token embeddings from the support tokens and input token embeddings from the input tokens. The second NER system may then recognize the new named entity in the input tokens using abstract tag transition probabilities and/or distances between the support token embeddings and the input token embeddings.

Abstract: Various embodiments provide for quantizing a trained neural network with removal of normalization with respect to at least one layer of the quantized neural network, such as a quantized multiple fan-in layer (e.g., element-wise add or sum layer).

Abstract: A device capable of autonomous motion may move in an environment and may receive audio data from a microphone. A model may be trained to process the audio data to suppress noise from the audio data. The model may include an encoder that includes one or more convolutional layers, one or more recurrent layers, and a decoder that includes one or more convolutional layers.

Abstract: In some example embodiments, there may be provided a method that includes receiving a machine learning model provided by a central server configured to provide federated learning; receiving first audio data obtained from at least one audio sensor monitoring at least one machine located at the first edge node; training, based on the first audio data, the machine learning model; providing parameter information to the central server in order to enable the federated learning among a plurality of edge nodes; receiving an aggregate machine learning model provided by the central server; detecting an anomalous state of the at least one machine. Related systems, methods, and articles of manufacture are also described.

Abstract: A method of processing an electrical signal transduced from a voice signal is disclosed. A classification model is applied to the electrical signal to produce a classification indicator. The classification model has been trained using an augmented training dataset. The electrical signal is classified as either one of a first class and a second class in a binary classification. The classifying being performed is a function of the classification indicator. A trigger signal is provided to a user circuit as a result of the electrical signal being classified in the first class of the binary classification.

Abstract: Techniques are described herein for detecting and/or enrolling (or commissioning) new “hot commands” that are usable to cause an automated assistant to perform responsive action(s) without having to be first explicitly invoked. In various implementations, an automated assistant may be transitioned from a limited listening state into a full speech recognition state in response to a trigger event. While in the full speech recognition state, the automated assistant may receive and perform speech recognition processing on a spoken command from a user to generate a textual command. The textual command may be determined to satisfy a frequency threshold in a corpus of textual commands. Consequently, data indicative of the textual command may be enrolled as a hot command. Subsequent utterance of another textual command that is semantically consistent with the textual command may trigger performance of a responsive action by the automated assistant, without requiring explicit invocation.