Abstract: Methods and apparatus for decoding a compressed Higher Order Ambisonics (HOA) representation of a sound or soundfield. The method may include receiving a bit stream containing the compressed HOA representation and decoding, based on a determination that there are multiple layers, the compressed HOA representation from the bitstream to obtain a sequence of decoded HOA representations. A first subset of the sequence of decoded HOA representations is determined based only on corresponding ambient HOA components. A second subset of the sequence of decoded HOA representations is determined based on corresponding ambient HOA components and corresponding predominant sound components.

Abstract: A method classifies feedback from transcripts. The method includes receiving an utterance from a transcript from a communication session and processing the utterance with a classifier model to identify a topic label for the utterance. The classifier model is trained to identify topic labels for training utterances. The topic labels correspond to topics of clusters of the training utterances. The training utterances are selected using attention values for the training utterances and clustered using encoder values for the utterances. The method further includes routing the communication session using the topic label for the utterance.

Abstract: A method and apparatus are provided for generating a personalized radio channel playlist by simultaneously buffering multiple received channels from one or more source streams, and then selecting songs or tracks to playback from the buffered channels. Users can specify favorite channels for building their personal playlists, or multiple default playlist channels can be provided by genre or channels related in some other way. Navigation tools permit users to skip ahead and backward in the playback stream. A personalized radio channel playlist can be implemented as (1) content selected from buffered channels based on user preferences for artists, songs and the like, or (2) as a Mix Channel in which content from selected buffered channels is automatically mixed for playback in response to selection of a preset button assigned to the Mix Channel.

Abstract: The purpose of the present invention is to provide an interactive system that allows addition of appropriate response content. An interactive system 100 includes a history DB 105b that stores search history information containing an acquired key, which is a keyword acquired from an input sentence acquired by user operation, a new candidate key, which is an unknown word, and a query sentence, which is response content retrieved using the acquired key. A query sentence generation unit 109 generates a new query sentence on the basis of the search history information, by using a search key contained in the query sentence and the new candidate key.

Abstract: An auxiliary device charging case is used to facilitate translation features of a mobile computing device or auxiliary device. A first user, who may be a foreign language speaker, holds the charging case and speaks into the charging case. The charging case communicates the received speech to the mobile computing device, either directly or through the auxiliary device, which translates the received speech into a second language for a second user, who is the owner of the mobile computing device and auxiliary device. The second user may provide input in the second language, such as by speaking or typing into the auxiliary or mobile computing device. The mobile computing device may translate this second input to the first language, and transmit the translated input to the charging case either directly or through the auxiliary device. The charging case may output the translated second input to the first user, such as through a speaker or display screen.

Abstract: An utterance section detection device which is capable of detecting an utterance section with high accuracy on the basis of whether or not an end of a speech section is an end of utterance.

Abstract: An apparatus includes one or more processors configured to receive orientation data and to select, based on the orientation data, a particular filter from among multiple filters. The one or more processors are configured to perform signal processing operations associated with three-dimensional (3D) sound data based on the particular filter.

Abstract: Apparatuses, systems, and methods are provided for parallel construction for question generation (QG) corresponding to a content item. Text of at least a portion of the content item may be extracted as a source language text, at least a portion of which may be translated to generate a parallel text having a primary language different from the source language. The source and primary language texts may be aligned to create an aligned source language text and an aligned primary language text. A QG strategy may be determined and at least one step to be performed on the aligned primary language text may be generated based at least in part upon the determined QG strategy. At least one parallel operation corresponding to the at least one step to be performed on the aligned primary language text may be constructed, and a source language question may be generated.

Abstract: A method implemented by a computing system comprises generating, by the computing system, a fingerprint comprising a plurality of bin samples associated with audio content. Each bin sample is specified within a frame of the fingerprint and is associated with one of a plurality of non-overlapping frequency ranges and a value indicative of a magnitude of energy associated with a corresponding frequency range. The computing system removes, from the fingerprint, a plurality of bin samples associated with a frequency sweep in the audio content.

Abstract: The present disclosure provides a method and an apparatus of verifying information based on a voice interaction, a device, and a computer storage medium, and relates to a field of artificial intelligence technology. The present disclosure is implemented to include: acquiring a text of a voice response of a user to a voice inquiry, wherein the voice inquiry is provided for verifying information with the user; and inputting each character of the text of the voice response and a phonetic information associated with the each character to a pre-trained semantic analysis model so as to obtain an user intention information and/or an information of an object to be verified output by the pre-trained semantic analysis model, wherein the user intention information includes a confirmation, a denial, an answer, or a question.

Abstract: An information providing method includes: acquiring, from one or more voice recognition devices, regional infection information indicating one or more infection alert levels and one or more regions associated with the one or more infection alert levels, the one or more infection alert levels being obtained by the one or more voice recognition devices analyzing a voice signal; calculating, based on the regional infection information, an infection risk value representing a magnitude of a risk of infection in each of the one or more regions; and generating output information in accordance with the infection risk value for each of the one or more regions.

Abstract: Provided is a technology of learning an acoustic model with a certain degree of accuracy of sound recognition within a short calculation period.

Abstract: A data processing device includes: a digital signal processor; at least one processor; and at least one memory device configured to store a plurality of instructions, which when executed by the at least one processor, cause the at least one processor to operate to: output a first determination result relating to a scene of content through use of sound data; select processing for the sound data by a first selection method based on the first determination result; determine an attribute of the content from among a plurality of attribute candidates; and select the processing by a second selection method, which is different from the first selection method, based on a determination result of the attribute, wherein the digital signal processor is configured to execute the processing selected by the at least one processor on the sound data.

Abstract: Detecting a replay attack on a voice biometrics system comprises receiving a speech signal; forming an autocorrelation of at least a part of the speech signal; and identifying that the received speech signal may result from a replay attack based on said autocorrelation. Identifying that the received speech signal may result from a replay attack may be achieved by: comparing the autocorrelation with a reference value; and identifying that the received speech signal may result from a replay attack based on a result of the comparison of the autocorrelation with the reference value, or by: supplying the autocorrelation to a neural network trained to distinguish autocorrelations formed from speech signals resulting from replay attacks from autocorrelations formed from speech signals not resulting from replay attacks.

Abstract: A computer-implemented method for training a neural end-to-end aspect based sentiment analysis (ABSA) system includes: inputting a batch of samples of a dataset into the neural end-to-end ABSA system, where the neural end-to-end ABSA system includes: a contextual language encoder configured to embed tokens with context; a first self-attention network configured to, based on an output of the contextual language encoder, detect an aspect term and provide a first output corresponding to the aspect term; and a second self-attention network configured to, based on the output of the contextual language encoder, detect the aspect term and provide a second output corresponding to the aspect term; and based on the inputted batch of samples and a consistency loss function, selectively adjusting weights of the neural end-to-end ABSA system based on consistent aspect term detection by the first self-attention network and the second self-attention network.

Abstract: Systems and methods for identifying a name are disclosed herein. In some embodiments, an apparatus may determine an attribute and/or attribute cluster. In some embodiments, an apparatus may determine a component word set as a function of an attribute and/or attribute cluster. In some embodiments, an apparatus may determine a candidate name by combining component words. In some embodiments, an apparatus may determine an intelligibility rating and/or an appeal rating for a candidate name.

Abstract: Methods and systems for videoconferencing include generating work quality metrics based on emotion recognition of an individual such as a call center agent. The work quality metrics allow for workforce optimization. One example method includes the steps of receiving a video including a sequence of images, detecting an individual in one or more of the images, locating feature reference points of the individual, aligning a virtual face mesh to the individual in one or more of the images based at least in part on the feature reference points, dynamically determining over the sequence of images at least one deformation of the virtual face mesh, determining that the at least one deformation refers to at least one facial emotion selected from a plurality of reference facial emotions, and generating quality metrics including at least one work quality parameter associated with the individual based on the at least one facial emotion.

Abstract: Systems and methods for podcast playback in a system including a playback device and a mobile device as a system controller are disclosed. In one embodiment, a playback system comprising a first playback device and a mobile device, the mobile device comprising computer-readable medium having stored thereon instructions executable to perform a method comprising capturing user input selecting an alarm function, capturing user input selecting a time for playing an alarm on the first playback device, capturing user input selecting a podcast channel, updating the graphical user interface to reflect the selected podcast channel, capturing user input specifying what order to play podcast episodes from the selected podcast channel, and starting playback of a first podcast episode on the first playback device according to the specified order to play podcast episodes by the previous user input and the selected time for playing an alarm.

Abstract: Example embodiments relate to methods and systems for playback of adaptive music corresponding to an athletic activity. A user input is received from a user selecting an existing song for audible playback to the user, the song comprising a plurality of audio layers including at least a first layer, a second layer, and a third layer. Augmented playback of the existing song to the user is initiated by audibly providing the first layer but not the second layer. Physical activity information derived from a sensor corresponding to a real-time physical activity level of a user is received. If the physical activity level of the user is above a first activity level threshold, the augmented playback of the existing song is continued by audibly providing the first layer and the second layer to the user.

Abstract: The present disclosure proposes a speech conversion scheme for non-parallel corpus training, to get rid of dependence on parallel text and resolve a technical problem that it is difficult to achieve speech conversion under conditions that resources and equipment are limited. A voice conversion system and a training method therefor are included. Compared with the prior art, according to the embodiments of the present disclosure: a trained speaker-independent automatic speech recognition model can be used for any source speaker, that is, the speaker is independent; and bottleneck features of audio are more abstract as compared with phonetic posteriorGram features, can reflect decoupling of spoken content and timbre of the speaker, and meanwhile are not closely bound with a phoneme class, and are not in a clear one-to-one correspondence relationship. In this way, a problem of inaccurate pronunciation caused by a recognition error in ASR is relieved to some extent.