Abstract: A merchandiser for storing refrigerated product for purchase by a customer is provided. The merchandiser includes a housing defining a cavity; an opening defined by the housing, the opening providing access to the cavity; a door coupled to the housing and movable between a first position and a second position, wherein the door covers the opening in the first position and the door is positioned away from the opening in the second position such that the cavity is accessible; a lock coupled to the door, the lock actuable between a locked position and an unlocked position, wherein the lock is configured to lock the door in the first position in the locked position; and a controller coupled to the lock, the controller configured to actuate the lock to the unlocked position in response to receiving an unlock signal from a remote device.

Abstract: A game system is disclosed that enables users to play games on their respective devices while viewing associated information on a separate screen, such as that of a movie theater. One of the methods includes maintaining, by a system running a game server associated with an electronic game, communications with one or more user devices each running a game client associated with the electronic game. The communications include client game information from each user device that identifies user interactions with the respective game client. Display data that includes a projection view to be provided for presentation on a display, separate from the user devices, is generated to be viewed by the users of the user devices. The display data is generated by a camera game client, controlled by the system, from a location in a game world of the electronic game. The projection view is presented on the display.

Abstract: Described herein are systems and methods for augmenting neural speech synthesis networks with low-dimensional trainable speaker embeddings in order to generate speech from different voices from a single model. As a starting point for multi-speaker experiments, improved single-speaker model embodiments, which may be referred to generally as Deep Voice 2 embodiments, were developed, as well as a post-processing neural vocoder for Tacotron (a neural character-to-spectrogram model). New techniques for multi-speaker speech synthesis were performed for both Deep Voice 2 and Tacotron embodiments on two multi-speaker TTS datasets—showing that neural text-to-speech systems can learn hundreds of unique voices from twenty-five minutes of audio per speaker.

Abstract: An ear-plug assembly presents audio content to an ear canal of a user. The audio content may be based in part on sound in a local area surrounding the user. The ear-plug assembly detects, via one or more acoustic sensors, sound in the area around the user. The sound waves travel through an aperture in a body of the ear-plug assembly and are propagated to a waveguide to the one or more acoustic sensors. The ear-plug assembly processes the detected sound data in a controller, which instructs a speaker assembly to present audio content based in part on the detected sound data. The detected sounds may be amplified, attenuated, filtered, and/or augmented when presented by the speaker assembly.

Abstract: Embodiments relate to obtaining head-related transfer function (HRTF) through performing simulation using images of a user's head. The geometry of the user's head is determined based in part on one or more images of the user's head. The simulation of sound propagation from an audio source to the user's head is performed based on the generated geometry. The geometry may be represented in a three-dimensional meshes or principal component analysis (PCA)-based where the user's head is represented as a combination of representative three-dimensional shapes of test subjects' heads.

Abstract: Embodiments of a production-quality text-to-speech (TTS) system constructed from deep neural networks are described. System embodiments comprise five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For embodiments of the segmentation model, phoneme boundary detection was performed with deep neural networks using Connectionist Temporal Classification (CTC) loss. For embodiments of the audio synthesis model, a variant of WaveNet was created that requires fewer parameters and trains faster than the original. By using a neural network for each component, system embodiments are simpler and more flexible than traditional TTS systems, where each component requires laborious feature engineering and extensive domain expertise. Inference with system embodiments may be performed faster than real time.

Abstract: An audio system for customizing sound fields for increased user privacy. A microphone array of a headset detects sounds from one or more sound sources in a local area of the headset. The audio system estimates array transfer functions (ATFs) associated with the sounds, and determines determining sound field reproduction filters for a loudspeaker array of the headset using the ATFs. The audio system presents audio content, via the loudspeaker array, based in part on the sound field reproduction filters. The presented audio content has a sound field that has a reduced amplitude in a first damped region of the local area that includes a first sound source of the one or more sound sources.

Abstract: Described herein are systems and methods for augmenting neural speech synthesis networks with low-dimensional trainable speaker embeddings in order to generate speech from different voices from a single model. As a starting point for multi-speaker experiments, improved single-speaker model embodiments, which may be referred to generally as Deep Voice 2 embodiments, were developed, as well as a post-processing neural vocoder for Tacotron (a neural character-to-spectrogram model). New techniques for multi-speaker speech synthesis were performed for both Deep Voice 2 and Tacotron embodiments on two multi-speaker TTS datasets—showing that neural text-to-speech systems can learn hundreds of unique voices from twenty-five minutes of audio per speaker.

Abstract: A multi-mode ring scanner (MMRS) has a ring unit for wearing on a finger. The MMRS optionally has a wrist unit coupled to the ring unit, such as via a cable. The MMRS optionally communicates wirelessly with a computing device. The ring unit has one or more scanners (such as an optical scanner or an RFID tag reader). The ring unit optionally has two paddle switches for activation by inward pressure from fingers adjacent to the finger. The two switches enable specifying operation of the MMRS in a plurality of modes and/or to communicate a plurality of information codes to the computing device. The computing device is optionally enabled to assign a function to each combination of activation of the two switches. A scanning system including the MMRS optionally provides feedback to a user based on feedback from a host processor.

Abstract: Embodiments of a production-quality text-to-speech (TTS) system constructed from deep neural networks are described. System embodiments comprise five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For embodiments of the segmentation model, phoneme boundary detection was performed with deep neural networks using Connectionist Temporal Classification (CTC) loss. For embodiments of the audio synthesis model, a variant of WaveNet was created that requires fewer parameters and trains faster than the original. By using a neural network for each component, system embodiments are simpler and more flexible than traditional TTS systems, where each component requires laborious feature engineering and extensive domain expertise. Inference with system embodiments may be performed faster than real time.

Abstract: A microphone can include a cover having a series of slits and a nest. The nest can be configured to receive a first diaphragm, a second diaphragm, and a PCB in a stacked arrangement, such that the PCB is positioned between the first diaphragm and the second diaphragm. Also, the first diaphragm can define a first plane, the second diaphragm can define a second plane, and the PCB can define a third plane and the first plane, the second plane, and the third plane can extend parallel to one another. The cover can also include slits having a first length and a second length, and the first length can be greater than the second length. The slits can extend both radially and axially.

Abstract: A system creates an audio profile. The audio profile may be stored in a database. For example, the audio profile may be securely stored in a database of a social network and associated with a user account. The audio profile may contain data describing the way in which the specific user hears and interprets sounds. Systems and applications which present sounds to the user may access the audio profile and modify the sounds presented to the user based on the data in the audio profile to enhance the audio experience for the user.

Abstract: An audio system for providing content to a user. The system includes a first and a second transducer assembly of a plurality of transducer assemblies, an acoustic sensor, and a controller. The first transducer assembly couples to a portion of an auricle of the user's ear and vibrates over a first range of frequencies based on a first set of audio instructions. The vibration causes the portion of the ear to create a first range of acoustic pressure waves. The second transducer assembly is configured to vibrate over a second range of frequencies to produce a second range of acoustic pressure waves based on a second set of audio instructions. The acoustic sensor detects acoustic pressure waves at an entrance of the ear. The controller generates the audio instructions based on audio content to be provided to the user and the detected acoustic pressure waves from the acoustic sensor.

Abstract: An ice merchandiser for storing bagged ice for purchase by a customer is provided. The ice merchandiser includes a payment system structured to enable a patron to facilitate a bagged ice product transaction at the ice merchandiser, and an access control system structured to selectively provide access to the ice merchandiser during the bagged ice product transaction based on the payment system providing an indication that payment for the ice product has been validated.

Abstract: In some embodiments, a system for regulating a digital door lock based on facial recognition includes a graphical user interface, an optical camera, communication circuitry, and processing circuitry configured to receive a signal corresponding to an image of the face of an individual captured by the optical camera. The processing circuitry may determine an identity of the individual based on the image of the face and determine whether or not to instruct the digital door lock to unlock or cause the interface to display instructions for the individual to access reserved goods or services based on the identity the individual. In any embodiments, a system may include a temperature sensor or other physiological parameter sensor to enable the processing circuitry to determine whether to unlock a door or provide access to goods or services based on the individual's health status, which may include a body temperature and/or other parameter.

Abstract: A wind chime has a suspension system for suspending a plurality of chimes, a central cord that suspends a clapper adjacent the chimes, and a silencer slidably mounted on the central cord. The silencer includes a plurality of chime apertures radially spaced from a center area adjacent an outer perimeter, each sized and shaped to receive one of the plurality of chimes. The central cord extends through the silencer to a clapper. An adjustable locking mechanism is mounted on the central cord between the silencer and the clapper, for locking the silencer in either a silenced position wherein the main body is positioned inside of the plurality of chimes with each of the chimes being positioned within one of the chime apertures, and a stored position wherein the main body is positioned above the chimes.

Abstract: The present disclosure features compounds such as those having the Formulae (I) and (II), which can increase cystic fibrosis transmembrane conductance regulator (CFTR) activity as measured in human bronchial epithelial (hBE) cells. The present disclosure also features methods of treating a condition associated with decreased CFTR activity or a condition associated with a dysfunction of proteostasis comprising administering to a subject an effective amount of a disclosed compound, such as a compound of Formula (I) or (II).

Abstract: Disclosed is an article for use with apparatus for heating smokable material to volatilize at least one component of the smokable material, the article including: a malleable container defining a cavity; a mass of smokable material in the cavity; and heating material that is heatable by penetration with a varying magnetic field to heat the smokable material. Also disclosed is a system including: such an apparatus; and an article for use with the apparatus, the article including a container defining a cavity, and a mass of smokable material in the cavity; wherein the apparatus includes a heating zone for receiving at least a portion of the article, and a magnetic field generator for generating a varying magnetic field to be used in heating the smokable material when the portion of the article is in the heating zone.

Abstract: An audio system presented herein includes a transducer array, a sensor array, and a controller. The transducer array presents audio content to a user. The controller controls the transducer array to adjust a level of tactile content imparted to the user via actuation of at least one transducer in the transducer array while presenting the audio content to the user. The audio system can be part of a headset.

Abstract: The disclosed computer-implemented method may include (1) receiving, at a first device, a first stream that includes first media data from a first media object, (2) receiving, at the first device, a second stream that includes second media data from a second media object, (3) mixing, at the first device, the first media data and the second media data into a third stream, (4) compiling, while mixing the third stream, a metadata stream that includes information enabling separation of the first media data and the second media data from the third stream, (5) transmitting, from the first device to a second device, the third stream, and (6) transmitting, from the first device to the second device, the metadata stream to enable the second device to separate the first media data and the second media data from the third stream. Various other methods, systems, and computer-readable media are also disclosed.