Abstract: A method includes receiving a sequence of input audio frames corresponding to an utterance captured by a user device, the utterance including a plurality of words. For each input audio frame, the method includes predicting, using a word boundary detection model configured receive the sequence of input audio frames as input, whether the input audio frame is a word boundary. The method includes batching the input audio frames into a plurality of batches based on the input audio frames predicted as word boundaries, wherein each batch includes a corresponding plurality of batched input audio frames. For each of the plurality of batches, the method includes processing, using a speech recognition model, the corresponding plurality of batched input audio frames in parallel to generate a speech recognition result.

Abstract: A device comprises a riser board and a riser bracket. The riser board has a connector that receives an expansion card. The riser bracket is coupled to the riser board, and includes a body portion and a latch. The body portion is coupled to the riser board such that the body portion is spaced apart from the riser board connector. The latch is rotatably coupled to the body portion, and rotates between a first position and a second position. When the latch is in the first position and the expansion card is received by the riser board connector, the expansion card is removable from the riser board connector. When the latch is in the second position and the expansion card is received by the riser board connector, the latch engages the expansion card such that the expansion card is not removable from the riser board connector.

Abstract: The techniques disclosed herein provide low power, efficient systems, devices and circuits that detect facial movements of a user of a Mixed Reality (MR) device. An example battery operated system includes a pulse driver that generates pulse signals at a set frequency, while operated below the battery voltage. An LC filter receives the pulse signals and generates a transient response based on a capacitance value associated with an antenna. The capacitance of the antenna varies responsive to facial movements of the user based on the varying distance of the antenna to the users skin. A coupling capacitor AC couples the output of the LC filter to other system components, where the AC coupled output of the LC filter includes a detected signal amplitude and a detected signal phase of the LC filter responsive to facial movements of the user.

Abstract: The present disclosure relates to an antibody or antigen-binding fragment thereof that specifically binds to a spike protein of SARS-CoV-2. The present disclosure also relates to a pharmaceutical composition, a method for treating and/or preventing diseases and/or disorders caused by a coronavirus in a subject in need thereof, and a method for detecting a coronavirus in a sample.

Abstract: The present invention relates to a method for inspecting a welding quality of an electrode tab-electrode lead welded portion of a pouch-type lithium secondary battery, and the method includes: recognizing welding traces of the welded portion by a vision inspection device; measuring a size of each of the recognized welding traces; and determining whether the welded portion has been weakly welded, excessively welded or normally welded by comparing the measured size with a reference value.

Abstract: A light-emitting pixel structure is described that may include a group of light-emitting diode structures, where each of the light-emitting diode structures is operable to emit light characterized by a different peak emission wavelength. The structures may also include a patterned light absorption barrier characterized by a group of openings in the barrier, where each of the openings permit a transmission of a portion of the light from one of the light-emitting diode structures through the barrier. The structures may further include a metasurface layer operable to change a direction of at least some of the light transmitted through the openings of the patterned light absorption barrier from the light-emitting diode structures.

Abstract: Metal gate cutting techniques for fin-like field effect transistors (FinFETs) are disclosed herein. An exemplary method includes receiving an integrated circuit (IC) device structure that includes a substrate, one or more fins disposed over the substrate, a plurality of gate structures disposed over the fins, a dielectric layer disposed between and adjacent to the gate structures, and a patterning layer disposed over the gate structures. The gate structures traverses the fins and includes first and second gate structures. The method further includes: forming an opening in the patterning layer to expose a portion of the first gate structure, a portion of the second gate structure, and a portion of the dielectric layer; and removing the exposed portion of the first gate structure, the exposed portion of the second gate structure, and the exposed portion of the dielectric layer.

Abstract: An epitaxial structure for a high-electron-mobility transistor includes a substrate, a nucleation layer, a buffer layered unit, a channel layer, and a barrier layer sequentially stacked on one another in such order. The buffer layered unit includes a plurality of p-i-n heterojunction stacks. Each of the p-i-n heterojunction stacks includes p-type, i-type, and n-type layers which are made of materials respectively represented by chemical formulas of AlxGa(1-x)N, AlyGa(1-y)N, and AlzGa(1-z)N. For each of the p-i-n heterojunction stacks, x decreases and z increases along a direction away from the nucleation layer, and y is consistent and ranges from 0 to 0.7.

Abstract: A method includes receiving a sequence of acoustic frames characterizing one or more utterances as input to a multilingual automated speech recognition (ASR) model. The method also includes generating a higher order feature representation for a corresponding acoustic frame. The method also includes generating a hidden representation based on a sequence of non-blank symbols output by a final softmax layer. The method also includes generating a probability distribution over possible speech recognition hypotheses based on the hidden representation generated by the prediction network at each of the plurality of output steps and the higher order feature representation generated by the encoder at each of the plurality of output steps. The method also includes predicting an end of utterance (EOU) token at an end of each utterance. The method also includes classifying each acoustic frame as either speech, initial silence, intermediate silence, or final silence.

Abstract: A micro electro-mechanical film speaker device is provided. A signal input terminal of a signal amplifier circuit is connected to an external input circuit. A second output terminal of the signal amplifier circuit is connected to a negative terminal of a micro electro-mechanical film speaker. A first terminal of a first high-pass filter is connected to a first output terminal of the signal amplifier circuit. A second terminal of the first high-pass filter is connected to a positive terminal of the micro electro-mechanical film speaker. A first terminal of a second high-pass filter is connected to a feedback terminal of the signal amplifier circuit. A second terminal of the second high-pass filter is connected to the positive terminal of the micro electro-mechanical film speaker.

Abstract: A system includes a package layer with microchannels to spread heat localized in the package at an electronic die. The microchannel is integrated onto or into the package layer. The microchannel has a hollow heat conducting material through which a fluid is to flow to spread the heat. The microchannel has a triangular cross-section or a trapezoidal cross-section. The microchannel can be sealed in the integration process to result in a closed heat pipe structure in which liquid flows through expansion and compression in response to heating and cooling, respectively.

Abstract: A cooling tower control method, used for controlling a cooling tower having at least one sensor, includes: receiving and processing a received sensor data; based on the received sensor data, timing training a water outlet temperature prediction model; receiving a target water outlet temperature; traverse searching a plurality of control parameter combinations meeting the target water outlet temperature; selecting an energy-saving target control parameter combination from the plurality of control parameter combinations meeting the target water outlet temperature; and controlling the cooling tower based on the target control parameter combination.

Abstract: A game controller configured for a mobile device. The game controller includes a connection mechanism, two control handles, and at least one first pad. The control handles are respectively coupled to two opposite sides of the connection mechanism. Each of the two control handles has an accommodation recess, and the two accommodation recesses of the two control handles are configured to accommodate the mobile device. The first pad is removably disposed on one of the two control handles and located in the accommodation recess of the one of the two control handles, and the at least one first pad is configured to be located between and clamped by the mobile device and the one of the two control handles.

Abstract: To provide a negative type photosensitive composition which is capable of forming a cured film having good light shielding properties and high reflectance. [Means for Solution] A negative type photosensitive composition comprising an alkali-soluble resin having a particular structure, a reflectance modifier, a polymerization initiator, and a solvent.

Abstract: A convertible notebook computer includes a metal mechanism element, a first radiation element, a second radiation element, a third radiation element, a first parasitic element, a second parasitic element, a third parasitic element, and a dielectric substrate. The metal mechanism element has a closed slot. The first radiation element has a feeding point. The second radiation element is coupled to the first radiation element. The third radiation element is coupled to the first radiation element. The first parasitic element is adjacent to the second radiation element. The second parasitic element is adjacent to the third radiation element. The third parasitic element is adj acent to the first radiation element. An antenna structure is formed by the closed slot of the metal mechanism element, the first radiation element, the second radiation element, the third radiation element, the first parasitic element, the second parasitic element, and the third parasitic element.

Abstract: A user-customized skin care method according to an embodiment of the present disclosure is capable of facilitating care for skin improvement by providing a user with customized skin care instructions through an analysis of a user skin condition by estimation models. In the customized skin care method, operation of a skin care device of a user is controlled according to skin care instructions estimated on the basis of personal information and skin condition information of the user, so that the user is provided with optimum customized skin care.

Abstract: The present disclosure generally relates to a system and a method of assessing the lungs of a patient using acoustic mapping to provide an image of the lung function and modifying therapy applied based on the information gathered.

Abstract: A method includes receiving training data comprising a plurality of pairs of images. Each pair comprises a noisy image and a denoised version of the noisy image. The method also includes training a multi-task diffusion model to perform a plurality of image-to-image translation tasks, wherein the training comprises iteratively generating a forward diffusion process by predicting, at each iteration in a sequence of iterations and based on a current noisy estimate of the denoised version of the noisy image, noise data for a next noisy estimate of the denoised version of the noisy image, updating, at each iteration, the current noisy estimate to the next noisy estimate by combining the current noisy estimate with the predicted noise data, and determining a reverse diffusion process by inverting the forward diffusion process to predict the denoised version of the noisy image. The method additionally includes providing the trained diffusion model.

Abstract: A display device includes a first substrate, a second substrate disposed on the first substrate, a display disposed on the first substrate, a sealant disposed between the first substrate and the second substrate along an outer portion of the second substrate and attaching the first substrate to the second substrate together, and a filler disposed in a space defined by the first substrate, the second substrate and the sealant. The filler is spaced apart from an inner surface of the sealant with an empty space disposed therebetween. An edge of the filler is inclined with respect to a first surface of the second substrate.

Abstract: An optical connector holding one or more optical ferrule assembly is provided. The optical connector includes an outer body, an inner front body accommodating the one or more optical ferrule assembly, ferrule springs for urging the optical ferrules towards a mating receptacle, and a back body for supporting the ferrule springs. The outer body and the inner front body are configured such that four optical ferrule assembly are accommodated in a small form-factor pluggable (SFP) transceiver footprint or eight optical ferrule assembly are accommodated in a quad small form-factor pluggable (QSFP) transceiver footprint. A receptacle can hold one or more connector inner bodies forming a single boot for all the optical fibers of the inner bodies.