An electronic apparatus is provided. The electronic apparatus includes a microphone, a transceiver, a memory configured to store a control command identification tool based on a control command identified by a voice recognition server that performs voice recognition processing on a user voice received from the electronic apparatus, and at least one processor configured to, based on the user voice being received through the microphone, acquire user intention information by performing the voice recognition processing on the received user voice, receive status information of external devices related to the acquired user intention information from a device control server, identify a control command for controlling a device to be controlled among the plurality of external devices by applying the acquired user intention information and the received status information of the external devices to the control command identification tool, and transmit the identified control command to the device control server.

Audio signals representing a current utterance in a conversation and a dialog history including at least information associated with past utterances corresponding to the current utterance in the conversation can be received. The dialog history can be encoded into an embedding. A spoken language understanding neural network model can be trained to perform a spoken language understanding task based on input features including at least speech features associated with the received audio signals and the embedding. An encoder can also be trained to encode a given dialog history into an embedding. The spoken language understanding task can include predicting a dialog action of an utterance. The spoken language understanding task can include predicting a dialog intent or overall topic of the conversation.

Audio signals representing a current utterance in a conversation and a dialog history including at least information associated with past utterances corresponding to the current utterance in the conversation can be received. The dialog history can be encoded into an embedding. A spoken language understanding neural network model can be trained to perform a spoken language understanding task based on input features including at least speech features associated with the received audio signals and the embedding. An encoder can also be trained to encode a given dialog history into an embedding. The spoken language understanding task can include predicting a dialog action of an utterance. The spoken language understanding task can include predicting a dialog intent or overall topic of the conversation.

Embodiments are associated with conditional teacher-student model training. A trained teacher model configured to perform a task may be accessed and an untrained student model may be created. A model training platform may provide training data labeled with ground truths to the teacher model to produce teacher posteriors representing the training data. When it is determined that a teacher posterior matches the associated ground truth label, the platform may conditionally use the teacher posterior to train the student model. When it is determined that a teacher posterior does not match the associated ground truth label, the platform may conditionally use the ground truth label to train the student model. The models might be associated with, for example, automatic speech recognition (e.g., in connection with domain adaptation and/or speaker adaptation).

Systems, methods, and devices are provided for generating and using text-to-speech (TTS) data for improved speech recognition models. A main model is trained with keyword independent baseline training data. In some instances, acoustic and language model sub-components of the main model are modified with new TTS training data. In some instances, the new TTS training is obtained from a multi-speaker neural TTS system for a keyword that is underrepresented in the baseline training data. In some instances, the new TTS training data is used for pronunciation learning and normalization of keyword dependent confidence scores in keyword spotting (KWS) applications. In some instances, the new TTS training data is used for rapid speaker adaptation in speech recognition models.

Techniques are described herein for improving performance of machine learning model(s) and thresholds utilized in determining whether automated assistant function(s) are to be initiated. A method includes: receiving, via one or more microphones of a client device, audio data that captures a spoken utterance of a user; processing the audio data using a machine learning model to generate a predicted output that indicates a probability of one or more hotwords being present in the audio data; determining that the predicted output satisfies a secondary threshold that is less indicative of the one or more hotwords being present in the audio data than is a primary threshold; in response to determining that the predicted output satisfies the secondary threshold, prompting the user to indicate whether or not the spoken utterance includes a hotword; receiving, from the user, a response to the prompting; and adjusting the primary threshold based on the response.

AM and LM parameters to be used for adapting an ASR model are derived for each audio segment of an audio stream comprising multiple audio programs. A set of identifiers, including a speaker identifier, a speaker domain identifier and a program domain identifier, is obtained for each audio segment. The set of identifiers are used to select most suitable AM and LM parameters for the particular audio segment. The embodiments enable provision of maximum constraints on the AMs and LMs and enable adaptation of the ASR model on the fly for audio streams of multiple audio programs, such as broadcast audio. This means that the embodiments enable selecting AM and LM parameters that are most suitable in terms of ASR performance for each audio segment.

A method for automatic hotword threshold tuning includes receiving, from a user device executing a first stage hotword detector configured to detect a hotword in streaming audio, audio data characterizing the detected hotword. The method includes processing, using a second stage hotword detector, the audio data to determine whether the hotword is detected by the second stage hotword detector. When the hotword is not detected, the method includes identifying a false acceptance instance at the first stage hotword detector indicating that the first stage hotword detector incorrectly detected the hotword. The method includes determining whether a false acceptance rate satisfies a false acceptance rate threshold based on a number of false acceptance instances within a false acceptance time period. When the false acceptance rate satisfies the false acceptance rate threshold, the method includes adjusting the hotword detection threshold of the first stage hotword detector.

A method for automatic hotword threshold tuning includes receiving, from a user device executing a first stage hotword detector configured to detect a hotword in streaming audio, audio data characterizing the detected hotword. The method includes processing, using a second stage hotword detector, the audio data to determine whether the hotword is detected by the second stage hotword detector. When the hotword is not detected, the method includes identifying a false acceptance instance at the first stage hotword detector indicating that the first stage hotword detector incorrectly detected the hotword. The method includes determining whether a false acceptance rate satisfies a false acceptance rate threshold based on a number of false acceptance instances within a false acceptance time period. When the false acceptance rate satisfies the false acceptance rate threshold, the method includes adjusting the hotword detection threshold of the first stage hotword detector.

A system for and method of characterizing a target application acoustic domain analyzes one or more speech data samples from the target application acoustic domain to determine one or more target acoustic characteristics, including a CODEC type and bit-rate associated with the speech data samples. The determined target acoustic characteristics may also include other aspects of the target speech data samples such as sampling frequency, active bandwidth, noise level, reverberation level, clipping level, and speaking rate. The determined target acoustic characteristics are stored in a memory as a target acoustic data profile. The data profile may be used to select and/or modify one or more out of domain speech samples based on the one or more target acoustic characteristics.