Abstract: A system for domain adaptation includes a domain adaptation model configured to adapt a representation of a signal in a first domain to a second domain to generate an adapted presentation and a plurality of discriminators corresponding to a plurality of bands of values of a domain variable. Each of the plurality of discriminators is configured to discriminate between the adapted representation and representations of one or more other signals in the second domain.

Abstract: An earpiece includes an earpiece housing, an intelligent control system disposed within the earpiece housing, a speaker operatively connected to the intelligent control system, a microphone operatively connected to the intelligent control system, and at least one sensor operatively connected to the intelligent control system for providing sensor data. The intelligent control system of the earpiece is configured to interface with a user of the earpiece by determining at least one of attention or intention of the user using the sensor data without receiving manual input at the earpiece and without receiving voice input from the user. The earpiece may be further configured to present an audio menu and use the attention or intention of the user to select one or more items from the audio menu.

Abstract: Disclosed are an apparatus and/or method for providing a sentence based on a user input. When a sentence corresponding to a shorthand word input from a user is generated and provided to the user, a sentence most suitable for a current context is provided to the user by further considering context information. At least a portion of the method for providing a sentence based on a user input may be performed using a rule-based model and/or an artificial intelligence model learned according to at least one of neural network or deep learning algorithms. The rule-based model and/or artificial intelligence model may provide a sentence most suitable for a current context to the user by using the input shorthand word and the context information as input values.

Abstract: Various examples are provided related to speech recognition. In one example, a method includes converting an auditory signal into a pulse train, segmenting the pulse train into a series of frames having a predefined duration, and identifying a portion of the auditory signal by applying at least a portion of the series of frames segmented from the pulse train to a kernel adaptive autoregressive-moving-average (KAARMA) network. In another example, a speech recognition system includes processing circuitry configured to convert an auditory signal into a pulse train, segment the pulse train into a secured of frames, and identifying a portion of the auditory signal by applying at least a portion of the series of frames segmented from the pulse train to a KAARMA network. The series of frames segmented from the pulse train can be applied to a KAARMA chain including a plurality of KAARMA networks for identification.

Abstract: A method, including obtaining data indicative of respective perceived loudness levels for a plurality of hearing percepts respectively evoked at different current levels, and creating a map for the hearing prosthesis based on the obtained data by adjusting at least one of the respective current levels based on data of a respective perceived loudness for another current level.

Abstract: Systems and methods for generating and annotating cell documents include extracting tables from a document using a table extraction engine. Headers are extracted for each of the tables using a header detection engine. Cells are extracted from each of the tables using a cell extraction engine. A cell document is generated for each of the cells which are each correlated to corresponding portions of the headers, each cell document recording the correlation between the cells and the the headers. Each cell document is annotated to generate annotated cell documents with a cell recognition model trained to perform natural language processing on the cell documents by classifying each term in each of the cell documents and extracting relationships between the terms of each of the cell documents.

Abstract: A mobile construction machine detects a speech processing trigger. It then performs speech processing (such as speech recognition and natural language understanding, speech synthesis, etc.) based on the detected speech processing trigger, to generate a speech processing result. A control signal generator generates control signals based on the speech processing result. The control signals can be used to control the mobile construction machine, to control another mobile construction machine, to provide information to a remote server location, or to aggregate information from multiple remote server locations.

Abstract: An earbud may have a housing with an ear portion and an elongated out-of-ear portion that protrudes away from the ear portion. A speaker may be aligned with a speaker port in the ear portion and may emit sound for a user. Audio playback functions and other operations may be controlled using a controller in the earbud. The controller may gather capacitive sensor data and other data and may use this data in identifying an operating mode of the earbud. Using information such as whether the earbud is in an in-ear state or an out-of-ear state or other sensor data, the controller may take actions such as pausing or resuming audio playback or adjusting playback volume. The capacitive sensor data can be gathered using capacitive sensing electrodes located on the ear portion and the stalk portion of the earbud.

Abstract: There is provided an information processing apparatus, an information processing method, and a program, in which a database in which unknown words are registered is able to be efficiently created, the information processing apparatus including: an identifying portion that identifies a word uttered according to a predetermined utterance method, from within utterance information of a user, as an unknown word; and a processing portion that performs processing to register the unknown word that has been identified.

Abstract: Techniques for receiving user input and soliciting responses from one or more users are described. A user may speak a question to a system. The system may determine that various question and answering content sources cannot provide an answer to the question. The system may then store data representing the question in a dedicated storage. The system may a topic, to which the question relates and provide the question to users having knowledge on the topic.

Abstract: The invention provides an efficient implementation of cross-product enhanced high-frequency reconstruction (HFR), wherein a new component at frequency Q?+r?0 is generated on the basis of existing components at ? and ?+?0. The invention provides a block-based harmonic transposition, wherein a time block of complex subband samples is processed with a common phase modification. Superposition of several modified samples has the net effect of limiting undesirable intermodulation products, thereby enabling a coarser frequency resolution and/or lower degree of oversampling to be used. In one embodiment, the invention further includes a window function suitable for use with block-based cross-product enhanced HFR. A hardware embodiment of the invention may include an analysis filter bank, a subband processing unit configurable by control data and a synthesis filter bank.

Abstract: An audio encoder can parse a digital audio signal into a plurality of frames, each frame including a specified number of audio samples, perform a transform of the audio samples of each frame to produce a plurality of frequency-domain coefficients for each frame, partition the plurality of frequency-domain coefficients for each frame into a plurality of bands for each frame, each band having a reshaping parameter that represents a time resolution and a frequency resolution, and encode the digital audio signal to a bit stream that includes the reshaping parameters. For a first band, the reshaping parameter can be encoded using a first alphabet size. For a second band, the reshaping parameter can be encoded using a second alphabet size different from the first alphabet size. Using different alphabet sizes can allow for more compact compression in the bit stream.

Abstract: A system and a method provide for protecting a loudspeaker from thermal and/or mechanical failure by monitoring for over-temperature and over-power conditions. The system generates a first gain from a first speaker protection controller in response to a driving voltage and/or a driving current of a loudspeaker, and generates a second gain from a second speaker protection controller in response to the driving voltage and/or a driving current of the loudspeaker, if the temperature exceeds a thermal limit or if the power exceeds a maximum power. The system applies the second gain to an audio signal to lower the audio signal if the first speaker protection controller fails.

Abstract: Disclosed is an apparatus and method for processing a multichannel audio signal. A multichannel audio signal processing method may include: generating an N-channel audio signal of N channels by down-mixing an M-channel audio signal of M channels; and generating a stereo audio signal by performing binaural rendering of the N-channel audio signal.

Abstract: The present disclosure provides methods, devices and computer program products for non-uniform quantization of parameters. The disclosure further relates to a method and apparatus for reconstructing an audio object in an audio decoding system taking the non-uniformly quantized parameters into account. According to the disclosure, such an approach renders it possible to reduce bit consumption without substantially reducing the quality of the reconstructed audio object.

Abstract: A system and method for spoken language understanding using recurrent neural networks (“RNNs”) is disclosed. The system and method jointly performs the following three functions when processing the word sequence of a user utterance: (1) classify a user's speech act into a dialogue act category, (2) identify a user's intent, and (3) extract semantic constituents from the word sequence. The system and method includes using a bidirectional RNN to convert a word sequence into a hidden state representation. By providing two different orderings of the word sequence, the bidirectional nature of the RNN improves the accuracy of performing the above-mentioned three functions. The system and method includes performing the three functions jointly. The system and method uses attention, which improves the efficiency and accuracy of the spoken language understanding system by focusing on certain parts of a word sequence. The three functions can be jointly trained, which increases efficiency.

Abstract: The various implementations described herein include methods, devices, and systems for recognizing speech, such as user commands. In one aspect, a method includes: (1) receiving audio input data via the one or more microphones; (2) generating a plurality of energy channels for the audio input data; (3) generating a feature vector by performing a per-channel normalization to each channel of the plurality of energy channels; and (4) obtaining recognized speech from the audio input utilizing the feature vector.

Abstract: A digital signal processor is programmed to extract a center channel from a stereo input, apply the center channel to an array of speaker elements using a first set of finite impulse response filters and a first rotation matrix to generate a first beam of audio content at a target angle about the axis, apply a left channel of the stereo input to the array of speaker elements using a second set of finite impulse response filters and a second rotation matrix to generate a second beam of audio content at a first offset angle from the target angle about the axis, and apply a right channel of the stereo input to the array of speaker elements using a third set of finite impulse response filters and third rotation matrix to generate a third beam of audio content at a second offset angle from the target angle about the axis.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to facilitating predicting style breaches within content. In one embodiment, target content for which style breach prediction is desired is obtained. Style features associated with the target content are identified. Such style features and a style breach prediction model are used to predict a style breach within the target content, the style breach indicating a change of style used within the target content (e.g., a single document).

Abstract: The present disclosure provides a speech recognition method and device. The method includes: receiving a speech signal; decoding the speech signal according to an acoustic model, a language model and a decoding network established in advance, and dynamically adding a blank unit in a decoding process to obtain an optimum decoding path with the added blank unit, in which the acoustic model is obtained based on connectionist temporal classification training, the acoustic model includes basic pronunciation units and the blank unit, and the decoding network includes a plurality of decoding paths consisting of the basic pronunciation units; and outputting the optimum decoding path as a recognition result of the speech signal.