Abstract: In an approach to AI explaining for natural language processing, responsive to receiving an input text for a machine learning model, an output is generated from the machine learning model. A plurality of alteration techniques are applied to the input text to generate one or more alternate outputs, where each alternate output corresponds to an alteration technique. A variation rate of the alternate output is calculated for each alteration technique. A preferred technique of generating neighboring data of the input text is generated based on a comparison of the variation rate of the alternate output for each alteration technique.

Abstract: Embodiments provide an audio processor for processing an audio signal to obtain a subband representation of the audio signal. The audio processor is configured to perform a cascaded lapped critically sampled transform on at least two partially overlapping blocks of samples of the audio signal, to obtain a set of subband samples on the basis of a first block of samples of the audio signal, and to obtain a corresponding set of subband samples on the basis of a second block of samples of the audio signal. Further, the audio processor is configured to perform a weighted combination of two corresponding sets of subband samples, one obtained on the basis of the first block of samples of the audio signal and one obtained on the basis on the second block of samples of the audio signal, to obtain an aliasing reduced subband representation of the audio signal.

Abstract: Hot-word free adaptation of one or more function(s) of an automated assistant. Sensor data, from one or more sensor components of an assistant device that provides an automated assistant interface (graphical and/or audible), is processed to determine occurrence and/or confidence metric(s) of various attributes of a user that is proximal to the assistant device. Whether to adapt each of one or more of the function(s) of the automated assistant is based on the occurrence and/or the confidence of one or more of the various attributes. For example, certain processing of at least some of the sensor data can be initiated, such as initiating previously dormant local processing of at least some of the sensor data and/or initiating transmission of at least some of the audio data to remote automated assistant component(s).

Abstract: An apparatus comprising means for: receiving values for sub-bands of a frame of an audio signal, the values comprising at least one azimuth value, at least one elevation value and at least one energy ratio value for each sub-band; determining an allocation of first number of bits to encode the values of the frame, wherein the first number of bits are fixed; encoding the at least one energy ratio value for a frame based on a defined allocation of a second number of bits from the first number of bits; encoding the at least one azimuth value and/or at least one elevation value of the frame based on a defined allocation of a third number of bits from the first number of bits, wherein the third number of bits is variably distributed on a sub-band-by-sub-band basis.

Abstract: An apparatus for processing an audio signal includes a configurable first audio signal processor for processing the audio signal in accordance with different configuration settings to obtain a processed audio signal, wherein the apparatus is adapted so that different configuration settings result in different sampling rates of the processed audio signal. The apparatus furthermore includes n analysis filter bank having a first number of analysis filter bank channels, a synthesis filter bank having a second number of synthesis filter bank channels, a second audio processor being adapted to receive and process an audio signal having a predetermined sampling rate, and a controller for controlling the first number of analysis filter bank channels or the second number of synthesis filter bank channels in accordance with a configuration setting.

Abstract: An electronic device comprising circuitry configured to perform a transcript based voice enhancement based on a transcript to obtain an enhanced audio signal.

Abstract: A method, a computing device, and a non-transitory machine-readable medium for detecting personal information. Terms that are of interest are extracted from a corpus of raw text that has been extracted from a collection of documents. For each of the terms, a surrounding sentence is extracted to form a target sentence to thereby form a plurality of target sentences. The surrounding sentence includes at least one reference to a data subject. A matrix of feature information is generated for each of the target sentences to form a plurality of matrices. A neural network model is trained, using the matrices as input, to compute an output that indicates a likelihood of a given sentence containing personal information.

Abstract: A computer receives a multimedia data, where the multimedia data comprises a plurality of frames. The computer converts the multimedia data into a signal wave having a plurality of frequencies and a plurality of amplitudes. The computer determines a frame from the plurality of frames having a pronoun. The computer identifies a topic of the frame. The computer searches for a frame in a media repository having a highest correlation coefficient with the topic of the frame, where the frame from the media repository comprises a bag of objects and resolves the anaphora disambiguation by substituting the pronoun with an object from the bag of objects.

Abstract: A method for managing electronic communication notifications includes responsive to receiving a communication from a first user, identifying one or more keywords in the communication based on a plurality of keywords associated with a plurality of queries previously presented by a second user. Determining whether the communication includes a reply to a first open query, wherein the first open query represents a question previously presented by the second user directed to the first user. Responsive to determining the communication from the first user includes the reply to the first open query, notifying the second user utilizing a first alert type for the communication from the first user that includes the reply for the first open query, wherein the first alert type is different from a second alert type for notifying the second user regarding a communication that does not include the reply for the first open query.

Abstract: A method and a system for generating medical reports based on medical images are provided. The method and system are characterized by carrying out an imaging examination on a target object; visualizing the image or the sequence of images acquired during the execution of the imaging examination; carrying out a report text generation step in parallel with the visualized image or sequence of images by using a speech recognition process; and saving the report text by associating univocally the report text to the visualized image or to the sequence of images.

Abstract: In some examples, a user, either a customer or potential customer of a business, engages in conversations with a virtual assistant (VA) provided by the business. The virtual assistant (VA) is further supported by one or more human assistants (HA), if needed. In embodiments, to facilitate seamless transitions between a VA and a HA, when needed, an intelligent decision maker (IDM) is provided. The IDM receives a user question and a proposed answer to the question from a VA, evaluates the proposed answer in the context of the conversation, and determines if the proposed answer requires further review by an HA. In response to a determination that the proposed answer requires further review, the IDM sends the proposed answer to an HA, and, in response to an indication by the HA, takes further action in the conversation.

Abstract: An audio processing method includes: converting a time-domain audio signal into a frequency-domain audio signal; determining a noise reduction gain according to the frequency-domain audio signal; and selecting at least one set of time-domain filter coefficients from a plurality sets of time-domain filter coefficients according to the noise reduction gain; configuring a time-domain filter according to the at least one selected set of time-domain filter coefficients, and filtering the time-domain audio signal with the time-domain filter.

Abstract: Disclosed is a machine learning-based media success prediction through plot summaries According to an embodiment, a method comprises performing preprocessing on text data including a plot summary, calculating a sentiment score from the preprocessed text data using a first model, generating first input data using the calculated sentiment score, generating second input data from the preprocessed data using a second model, and determining a candidate class of content corresponding to the plot summary by applying the first input data and the second input data to a pre-trained third model. The candidate class includes a first class indicating success and a second class indicating failure.

Abstract: Disclosed herein are systems, computer-implemented methods, and tangible computer-readable storage media for speaker recognition personalization. The method recognizes speech received from a speaker interacting with a speech interface using a set of allocated resources, the set of allocated resources including bandwidth, processor time, memory, and storage. The method records metrics associated with the recognized speech, and after recording the metrics, modifies at least one of the allocated resources in the set of allocated resources commensurate with the recorded metrics. The method recognizes additional speech from the speaker using the modified set of allocated resources. Metrics can include a speech recognition confidence score, processing speed, dialog behavior, requests for repeats, negative responses to confirmations, and task completions.

Abstract: A dispatcher virtual assistant (DVA) that can augment the capability of emergency dispatchers while reducing human errors. Major functions of the DVA include updating an emergency incident's status in real time, recommending or reminding the dispatcher to take proper actions at the right timing, answering the dispatcher's inquiries for task-related information, and fulfilling the dispatcher's request for an incident report. The DVA system includes a dispatcher language model based on machine-learning and deep-learning algorithms, for extracting the status of a live incident from incoming incident logs, and for processing and answering inquiries or requests from the dispatcher. It is customizable for different types of emergencies and for different local communities. The DVA can be used in tandem with an existing CAD system.

Abstract: Described embodiments include an apparatus that includes a network interface and a processor. The processor is configured to receive, via the network interface, a speech signal that represents speech uttered by a subject, the speech including one or more speech segments, divide the speech signal into multiple frames, such that one or more sequences of the frames represent the speech segments, respectively, compute respective estimated total volumes of air exhaled by the subject while the speech segments were uttered, by, for each of the sequences, computing respective estimated flow rates of air exhaled by the subject during the frames belonging to the sequence and, based on the estimated flow rates, computing a respective one of the estimated total volumes of air, and, in response to the estimated total volumes of air, generate an alert. Other embodiments are also described.

Abstract: A method for providing a natural language interface for a computer-aided design (CAD) system includes receiving a user voice input comprising a plurality of words, parsing the user voice input, determining a meaning for the parsed user voice input, the meaning including one or more words associated with an object and one or more words associated with a characteristic of the object, retrieving from a model descriptor database at least an object model descriptor and at least a characteristic descriptor, using the determined meaning, generating at least a graphical model of the object using the at least an object model descriptor, and generating at least a modified graphical model of the object, using the at least a characteristic descriptor.

Abstract: The speech processing apparatus 100 includes an air microphone speech recognition unit 101 which recognizes speech from an air microphone 200 acquiring speech through air, a wearable microphone speech recognition unit 102 which recognizes speech from a wearable microphone 300, a sensing unit 103 which measures environmental conditions, a weight decision unit 104 which calculates the weights for recognition results of the air microphone speech recognition unit 101 and the wearable microphone speech recognition unit 102 on the basis of the environmental conditions, and a combination unit 105 which combines the recognition results outputted from the air microphone speech recognition unit 101 and the wearable microphone speech recognition unit 102, using the weights.

Abstract: An illustrative embodiment includes a method for training an end-to-end (E2E) spoken language understanding (SLU) system. The method includes receiving a training corpus comprising a set of text classified using one or more sets of semantic labels but unpaired with speech and using the set of unpaired text to train the E2E SLU system to classify speech using at least one of the one or more sets of semantic labels. The method may include training a text-to-intent model using the set of unpaired text; and training a speech-to-intent model using the text-to-intent model. Alternatively or additionally, the method may include using a text-to-speech (TTS) system to generate synthetic speech from the unpaired text; and training the E2E SLU system using the synthetic speech.

Abstract: Embodiments described herein provide a dynamic topic tracking mechanism that tracks how the conversation topics change from one utterance to another and use the tracking information to rank candidate responses. A pre-trained language model may be used for response selection in the multi-party conversations, which consists of two steps: (1) a topic-based pre-training to embed topic information into the language model with self-supervised learning, and (2) a multi-task learning on the pretrained model by jointly training response selection and dynamic topic prediction and disentanglement tasks.