Abstract: Systems and processes for operating an intelligent automated assistant are provided. For example, a digital assistant obtains, from a software application, a first static vocabulary entry for the software application; registers the first vocabulary entry with a knowledge base for a digital assistant of the electronic device; and, while the software application is running: receives a request from the software application to register a second dynamic vocabulary entry for the software application; and registers the second vocabulary entry with the knowledge base for the digital assistant.

Abstract: Systems and methods are provided for generating a pseudo-labeled training dataset by at least one of: (1) extracting a set of intermediate outputs from an automatic speech recognition model based on applying the automatic speech recognition model to the set of unlabeled speech data, clustering the set of intermediate outputs into different clusters, and generating a first set of pseudo-labels comprising cluster assignments associated with the different clusters and which correspond to the unlabeled speech data, or (2) generating a set of decoded word sequences for the unlabeled speech data by applying the automatic speech recognition model to the set of unlabeled speech data, and generating a second set of pseudo-labels associated with the unlabeled speech data by applying the automatic speech recognition model to both (i) the set of decoded word sequences and (ii) the set of unlabeled speech data.

Abstract: Disclosed is a wireless display system comprising a wireless set-top box for communicating with a wireless display device and performing voice recognition, the wireless set-top box comprising: a communication unit that transmits image data and sound data to the wireless display device; a microphone that receives a voice command of a user; and a control unit that, when the voice command of the user is received, obtains an echo removal parameter on the basis of location information of the wireless set-top box and performs echo removal by using the echo removal parameter.

Abstract: An example process includes: receiving an audio stream; and while receiving the audio stream: in accordance with a determination, based on a semantic analysis of at least a portion of the audio stream, that a user request in the at least a portion of the audio stream is complete: providing a first response to the user request; and in accordance with a determination, based on the semantic analysis of the least a portion of the audio stream, that the user request is incomplete: forgoing providing, for a predetermined duration, an audible response to the user request.

Abstract: An electronic device includes a microphone; at least one memory storing a wake-up word detection model; and at least one processor configured to: obtain a sound signal received through the microphone, input the sound signal into the wake-up word detection model, obtain, as an output of the wake-up word detection model, one or more first probability scores corresponding to one or more sections of the sound signal, wherein each first probability score of the one or more first probability scores represents a probability that a corresponding section of the one or more sections of the sound signal corresponds to a wake-up word, identify a first section of the sound signal, among the one or more sections of the sound signal, that corresponds to a first probability score, among the one or more first probability scores, that exceeds a first threshold value, and based on identifying a predetermined acoustic signal in the sound signal, reduce the first threshold value.

Abstract: A non-transitory, processor-readable medium stores instructions that, when executed by a processor, cause the processor to provide prompt data as input to a first machine learning model to produce a plurality of training phrases. The plurality of training phrases is provided as input to a second machine learning model to produce a plurality of embeddings. Each embedding from the plurality of embeddings is provided as input to a third machine learning model to identify a classification associated with that embedding, each classification being from a plurality of classifications that defines a sequence. The instructions further cause the processor to train a fourth machine learning model to predict, based on text data received from a user compute device, the classification from the plurality of classifications.

Abstract: An apparatus comprising means for: classifying an audio signal as at least a first audio signal for time synchronization with a video signal or a second audio signal not for time synchronization with a video signal; processing the first audio signal for time synchronization with the video signal wherein the processing introduces a first delay for time synchronization with the video signal; processing the second audio signal wherein the processing introduces a second delay shorter than the first delay.

Abstract: Embodiments of the present disclosure provide a method, apparatus, electronic device, and storage medium for text content matching. The method of text content matching includes: in accordance with a collection of to-be-processed speech information, determining a to-be-processed acoustic feature corresponding to the to-be-processed speech information (S110); processing, based on an audio following method, the to-be-processed acoustic feature to obtain a to-be-matched utterance corresponding to the to-be-processed acoustic feature (S120); and determining a target utterance associated with the to-be-matched utterance in target text and differentiating a display of the target utterance in the target text (S130).

Abstract: A method includes transmitting a prompt to a large language model (LLM). The prompt includes an instruction to generate a first reasoning block. The method further includes computing a first token-level confidence score for the first reasoning block. The method further includes generating, by the LLM, a second reasoning block responsive to the first token-level confidence score failing to satisfy a pruning threshold. The method further includes computing a second token-level confidence score for the second reasoning block. Responsive to the second token-level confidence score satisfying the pruning threshold, the second reasoning block is added to a reasoning trace. The method further includes generating a provisional answer from the reasoning trace, responsive to the second token-level confidence score satisfying a block confidence threshold.

Abstract: A dialogue management system applies hierarchical classifiers and other natural language processing to dialogue input, and determines whether performance of an action is likely to occur. The dialogue management system may process dialogue input to assess a dialogue participant's current position in various hierarchies or other classification schemes associated with performance of a desired action. The system may then present results of the assessment to another dialogue participant or provide the results to another system. In some embodiments, the dialogue management system may automatically generate responses or questions designed to engage a dialogue participant and cause the participant to progress through the levels of a hierarchy toward performance of a desired action.

Abstract: An electronic apparatus includes a communication interface configured to communicate with an external device; a speaker; a memory storing at least one instruction; and at least one processor operatively connected to the communication interface, the speaker, and the memory, wherein the at least one processor is configured to execute the at least one instruction to: based on a voice signal being received from an external device through the communication interface, identify whether the external device is a privacy device; based on identifying that the external device is a privacy device, output a response signal regarding the voice signal to the external device through the communication interface; and based on identifying that the external device is not a privacy device, output a response signal regarding the voice signal through the speaker.

Abstract: A vehicular recording control device includes: an imaging data acquiring unit configured to acquire first imaging data from a first camera imaging surroundings of a host vehicle and second imaging data from a second camera imaging a cabin of the vehicle; a sightline determining unit configured to determine the direction of a sightline of an occupant of the vehicle from the second imaging data; a voice recognizing unit configured to recognize a voice command for instructing event recording; an output control unit configured to output information indicating the voice command to the occupant when the sightline determining unit determines that the sightline of the occupant is directed to a display unit; and a recording control unit configured to store the first imaging data as event data when the voice recognizing unit recognizes the voice command.

Abstract: A method for detecting short-lived repeat voice commands includes activating a set of one or more warm words each associated with a respective action for controlling a long-standing operation performed by a digital assistant. While the digital assistant is performing the long-standing operation, the method includes receiving audio data, and detecting, in the audio data, a warm word from the activated set of one or more warm words. In response to detecting the warm word, the method includes performing the respective action associated with the detected warm word, and activating a set of one or more repeat warm words associated with the detected warm word. The method further includes receiving additional audio data, detecting, in the additional audio data, a repeat warm word from the activated set of one or more repeat warm words, and performing the respective action associated with the detected repeat warm word.

Abstract: An electronic apparatus includes a communication interface configured to communicate with an external device; a speaker; a memory storing at least one instruction; and at least one processor operatively connected to the communication interface, the speaker, and the memory, wherein the at least one processor is configured to execute the at least one instruction to: based on a voice signal being received from an external device through the communication interface, identify whether the external device is a privacy device; based on identifying that the external device is a privacy device, output a response signal regarding the voice signal to the external device through the communication interface; and based on identifying that the external device is not a privacy device, output a response signal regarding the voice signal through the speaker.

Abstract: Implementations are directed to generating voice-based chatbot policy override(s) and/or utilizing voice-based chatbot policy override(s) in conjunction with existing voice-based chatbot(s). The voice-based chatbot policy override(s) can correspond to, for example, machine learning (ML) model(s) that supplement functionality of the existing voice-based chatbot(s). Notably, the voice-based chatbot policy override(s) are associated with rule(s) (e.g., by virtue of training the ML model(s) that correspond to the voice-based chatbot policy override(s)) for when the voice-based chatbot policy override(s) should be utilized in lieu of the existing voice-based chatbot(s) in responding to spoken utterance(s) of human user(s) engaged in corresponding conversation(s) with the voice-based chatbot policy override(s). Nonetheless, from a perspective of the human user(s), it appears as if they are still engaging in the corresponding conversations with the existing voice-based chatbot(s).

Abstract: In one aspect, a playback device includes a voice assistant service (VAS) wake-word engine and a command keyword engine. The playback device detects, via the command keyword engine, a first command keyword, and determines whether one or more playback conditions corresponding to the first command keyword are satisfied. Based on (a) detecting the first command keyword and (b) determining that the one or more playback conditions corresponding to the first command keyword are satisfied, the playback device playback device performs a first playback command corresponding to the first command keyword. When the playback device detects, via the wake-word engine, a wake-word in voice input, the playback device streams sound data corresponding to at least a portion of the voice input to one or more remote servers associated with the VAS.

Abstract: A sound collection apparatus A server (10) causes a terminal device (30) to display a setting user interface, and receives a prohibition time zone setting command. Server sets a prohibition time zone for a target device based on the prohibition time zone setting command. Server (10) causes the target device to utter, when a time for causing the target device to utter is determined not to correspond to the prohibition time zone.

Abstract: In some embodiments, the present invention provides for an exemplary computer system which includes at least the following components: an adaptive self-trained computer engine programmed, during a training stage, to electronically receive an initial speech audio data generated by a microphone of a computing device; dynamically segment the initial speech audio data and the corresponding initial text into a plurality of user phonemes; dynamically associate a plurality of first timestamps with the plurality of user-specific subject-specific phonemes; and, during a transcription stage, electronically receive to-be-transcribed speech audio data of at least one user; dynamically split the to-be transcribed speech audio data into a plurality of to-be-transcribed speech audio segments; dynamically assigning each timestamped to-be-transcribed speech audio segment to a particular core of the multi-core processor; and dynamically transcribing, in parallel, the plurality of timestamped to-be-transcribed speech audio segm

Abstract: Systems and methods to provide action recommendations based on audio information are disclosed.

Abstract: Dialog acts (e.g., questions) are selected for voice browsing by a model trained to identify a dialog act that is most likely to lead to a desired outcome. Upon receiving an invocation to begin a conversation, a score indicative of a level of confidence that the conversation will have a successful outcome is determined, and a dialog act is selected based on the score. Subsequently, at each turn of the conversation, the score is updated or a new score is calculated, and a dialog act is selected based on the updated or new score. Confidence scores are calculated based on input features that are determined based on the user who uttered the invocation or responses to dialog acts, as well as a context of the conversation, and provided to a linear model or a machine learning model as inputs.