Abstract: A method for training parameters of a first domain adaptation model includes evaluating a cycle consistency objective using a first task specific model associated with a first domain and a second task specific model associated with a second domain. The evaluating the cycle consistency objective is based on one or more first training representations adapted from the first domain to the second domain by a first domain adaptation model and from the second domain to the first domain by a second domain adaptation model, and one or more second training representations adapted from the second domain to the first domain by the second domain adaptation model and from the first domain to the second domain by the first domain adaptation model. The method further includes evaluating a learning objective based on the cycle consistency objective, and updating parameters of the first domain adaptation model based on learning objective.

Abstract: Methods and systems for enhanced natural language processing of clinical documentation are provided. Using natural language processing, a clinical condition is extracted from unstructured data within a current electronic document. A clinical ontology identifying itemsets associated with the clinical condition is retrieved, and indicators of relevant clinical concepts, as identified from the ontology, are searched from within the patient's longitudinal record, which comprises documentation from at least a prior encounter. Based on the whether the clinical concepts are present in the patent's record, a confidence is assigned to the NLP-extracted clinical condition, and one or more actions may be performed.

Abstract: An electronic device is disclosed. The electronic device comprises: multiple microphones for receiving audio signals generated by multiple sound sources; a communication unit for communicating with a voice recognition server; and a processor for determining the direction in which each of the multiple sound sources is located with reference to the electronic device, on the basis of the multiple audio signals received through the multiple microphones, determining at least one target sound source among the multiple sound sources on the basis of the duration of the determined direction of each of the sound sources, and controlling the communication unit such that the communication unit transmits, to the voice recognition server, an audio signal of a target sound source from which a predetermined voice is generated among the at least one target sound source.

Abstract: A speech interface device is configured to receive response data from a remote speech processing system for responding to user speech. This response data may be enhanced with information such as remote NLU data. The response data from the remote speech processing system may be compared to local NLU data to improve a speech processing model on the device. Thus, the device may perform supervised on-device learning based on the remote NLU data. The device may determine differences between the updated speech processing model and an original speech processing model received from the remote system and may send data indicating these differences to the remote system. The remote system may aggregate data received from a plurality of devices and may generate an improved speech processing model.

Abstract: An example operation may include one or more of receiving a plurality of local audio files from a plurality of audio devices that participated in a teleconference, where each local audio file includes a locally captured audio recording of a user of a respective audio device during the teleconference, generating combined audio playback information for the teleconference based on the plurality of local audio files received from the plurality of audio devices, the generating including detecting audio portions within the plurality of local audio files and synchronizing a playing order of the detected audio portions based on timing information included in the plurality of local audio files, and transmitting the combined audio playback information of the teleconference to at least one audio device among the plurality of audio devices.

Abstract: Devices and techniques are generally described for implementing a vehicle voice user interface. In some examples, a first request may be received from a first computing system. In some further examples, natural language data comprising a semantic representation of the first request may be determined. In some further examples, first metadata corresponding to the first computing system may be determined. In some examples, a restriction on execution of the first action may be determined in accordance with the first metadata. In at least some further examples, output data indicative of the restriction on execution of the first action may be sent to the first computing system.

Abstract: Computer-implemented methods, computer systems and computer program products for providing geographic location specific models for information extraction and knowledge discovery are provided. Aspects include receiving a body of input text using a processor having natural language processing functionality. Aspects also include using information extraction functionality of the processor to extract preliminary information including a relational table from the body of input text. Aspects also include determining one or more geographical contexts associated with the input text based on the preliminary information. Aspects also include determining inferred information based on the preliminary information and the one or more geographical contexts associated with the input text. Aspect also include augmenting the relational table with the inferred information.

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: An application processor is provided. The application processor includes a system bus, a host processor, a voice trigger system and an audio subsystem that are electrically connected to the system bus. The voice trigger system performs a voice trigger operation and issues a trigger event based on a trigger input signal that is provided through a trigger interface. The audio subsystem processes audio streams through an audio interface. While an audio replay is performed through the audio interface, the application processor performs an echo cancellation with respect to microphone data received from a microphone to generate compensated data and the voice trigger system performs the voice trigger operation based on the compensated data.

Abstract: A method of attention-based end-to-end (A-E2E) automatic speech recognition (ASR) training, includes performing cross-entropy training of a model, based on one or more input features of a speech signal, determining a posterior probability vector at a time of a first wrong token among one or more output tokens of the model of which the cross-entropy training is performed, and determining a loss of the first wrong token at the time, based on the determined posterior probability vector. The method further includes determining a total loss of a training set of the model of which the cross-entropy training is performed, based on the determined loss of the first wrong token, and updating the model of which the cross-entropy training is performed, based on the determined total loss of the training set.

Abstract: The present invention relates to an electronic device and a control method thereof. The electronic device comprises a microphone for acquiring sound; and a control unit for determining whether the acquired sound is a learned sound and outputting information on the acquired sound on the basis of a determination result.

Abstract: A computer-implemented method includes obtaining a training data set including text data indicating one or more phrases or sentences. The computer-implemented method includes training a classifier using supervised machine learning based on the training data set and additional text data indicating one or more out-of-domain phrases or sentences. The computer-implemented method includes training an autoencoder using unsupervised machine learning based on the training data. The computer-implemented method further includes combining the classifier and the autoencoder to generate the out-of-domain sentence detector configured to generate an output indicating a classification of whether input text data corresponds to an out-of-domain sentence. The output is based on a combination of a first output of the classifier and a second output of the autoencoder.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for coding speech using neural networks. One of the methods includes obtaining a bitstream of parametric coder parameters characterizing spoken speech; generating, from the parametric coder parameters, a conditioning sequence; generating a reconstruction of the spoken speech that includes a respective speech sample at each of a plurality of decoder time steps, comprising, at each decoder time step: processing a current reconstruction sequence using an auto-regressive generative neural network, wherein the auto-regressive generative neural network is configured to process the current reconstruction to compute a score distribution over possible speech sample values, and wherein the processing comprises conditioning the auto-regressive generative neural network on at least a portion of the conditioning sequence; and sampling a speech sample from the possible speech sample values.

Abstract: A dialogue system for a vehicle may include: an input processor configured to receive a dialogue among occupants of the vehicle including a driver and at least one passenger, to detect vehicle operation information, to identify the at least one passenger based on the dialogue among the occupants or the vehicle operation information, to generate passenger number information which estimates a change in a number of passengers in the vehicle based on the dialogue among the occupants when the vehicle arrives at a stop-over point, and to acquire a pre-utterance message according to the passenger number information; and a result processor configured to output a pre-utterance according to the pre-utterance message.

Abstract: A dialogue system for a vehicle may include: a storage storing context information including at least one of vehicle state information indicating a vehicle state, user information indicating a driver of the vehicle, or driving environment information indicating a driving environment of the vehicle; an input processor determining whether a context of the vehicle is a pre-utterance context, in which the dialogue system firstly outputs an utterance, based on the stored context information, determining whether a task related to the pre-utterance context is a duplicate task which has already processed or is currently processing when the context of the vehicle is the pre-utterance context, and obtaining at least one of an action corresponding to the pre-utterance context or a pre-utterance message to be output by the dialogue system when the task related to the pre-utterance context is not a duplicate task; a dialogue manager obtaining, from the storage, a parameter value of an action parameter used to perform the

Abstract: An Internet of Thing (IoT) device includes a transceiver coupled to a processor. Contracts can be used with the device to facilitate secure operation.

Abstract: Systems and methods are disclosed for audio enhancement. For example, methods may include accessing audio data; determining a window of audio samples based on the audio data; inputting the window of audio samples to a classifier to obtain a classification, in which the classifier includes a neural network and the classification takes a value from a set of multiple classes of audio; selecting, based on the classification, an audio enhancement network from a set of multiple audio enhancement networks; applying the selected audio enhancement network to the window of audio samples to obtain an enhanced audio segment, in which the selected audio enhancement network includes a neural network that has been trained using audio signals of a type associated with the classification; and storing, playing, or transmitting an enhanced audio signal based on the enhanced audio segment.

Abstract: A conversation may be monitored in real time using a trained machine learning model to identify a desired outcome of a conversation and generate one or more phrases for accomplishing the desired outcome. A confidence score may also be determined for one or more phrases that indicates a likelihood that the one or more phrases may help accomplish the desired outcome of the conversation. In some examples, a confidence score may be based on whether an agent, a caller, or both responded unfavorably to a similar phrase used previously in another conversation. In other examples, a confidence score corresponding to one or more phrases may be based on whether a prior conversation in which one or more similar phrases was used resulted in the desired outcome being accomplished.

Abstract: One embodiment provides a method, including: detecting, at an information handling device, voice command input; determining, using a processor, an orientation state of the information handling device, wherein the orientation state is one of: a closed state or an open state; and displaying, responsive to determining that the orientation state is associated with the closed state and on an auxiliary display of the information handling device, a notification acknowledging receipt of the voice command input. Other aspects are described and claimed.

Abstract: Performing an operation comprising analyzing an electronic text corpus by a natural language processing (NLP) algorithm executing on a processor to determine a set of target terms specified in the electronic text corpus, analyzing the electronic text corpus by the NLP algorithm to determine a weight for each pair of terms in the set of target terms, and storing an indication of each pair of terms and the respective weight in a lookup table in a computer memory for runtime analysis of an input text data.