Abstract: An event-guided auto-exposure (EG-AE) device leverages the high dynamic range and low latency properties of a dynamic vision sensor to guide the exposure setting of an active pixel sensor. The method utilizes physical connections of images and events to estimate the irradiance variations, which are fed into the EG-AE device for calculating the exposure setting. The method shows outstanding performance in tackling highly challenging lighting conditions and benefits multiple downstream applications, including visual tag detection and feature matching.

Abstract: An audio synthesis method, an electronic device, and a non-transitory computer readable storage medium. The method includes: in response to obtaining a text to be processed, inserting a colloquial expression content into the text to be processed and obtaining a colloquialized text, where the colloquial expression content at least includes a colloquial additional content and a colloquial pause interval; acquiring a prosodic feature of the colloquialized text and obtaining a prosodic pause interval of the colloquialized text based on the prosodic feature; and converting the colloquialized text into a target audio based on the colloquial pause interval and the prosodic pause interval.

Abstract: The present disclosure provides a semiconductor structure and a method of forming the same. The semiconductor structure includes: a substrate doped with a first ion, a deep trench structure disposed in the substrate, a barrier doped region disposed on a top of the substrate and the deep trench structure, a first epitaxial layer disposed on the barrier doped region, a body region disposed in the first epitaxial layer, a source region disposed in the body region, a gate structure disposed in the first epitaxial layer, and a collector region disposed at a bottom of the substrate. By means of the semiconductor structure, performance of an insulated gate bipolar transistor can be improved.

Abstract: A speech recognition method is provided. The method includes: obtaining a to-be-recognized speech and a speech recognition model, including an encoding network and a decoding network, after training; during each stage of encoding the to-be-recognized speech using the encoding network, classifying the to-be-recognized speech under a target speech attribute to obtain a predicted attribute category, and performing encoding to obtain a first encoding feature according to the predicted attribute category under the target speech attribute; decoding the first encoding feature according to the decoding network to obtain a recognition text of the to-be-recognized speech, the speech recognition model being adjusted according to at least a first loss, which represents a difference between a preset attribute category annotated in a speech sample and a sample attribute category recognized and obtained by the speech recognition model under the target speech attribute.

Abstract: A speech recognition error correction method and device, and a readable storage medium are provided. The method includes: acquiring to-be-recognized speech data and a first recognition result of the speech data, re-recognizing the speech data with reference to context information in the first recognition result to obtain a second recognition result, and determining a final recognition result based on the second recognition result. In the method, the speech data is re-recognized with reference to context information in the first recognition result, which fully considers context information in the recognition result and the application scenario of the speech data. If any error occurs in the first recognition result, the first recognition result is corrected based on the second recognition. Therefore, the accuracy of speech recognition can be improved.

Abstract: A smart walking assistance device with a walker frame having generally vertical sides and an intersecting front. Wheels located at both ends of a bottom edge of the sides. A soft robotic sensing handle extends in a C shape along the upper edges of the sides and front. The sensing handle has multiple contiguous air filled chambers, each containing a pressure sensor for producing a pressure signal representing the pressure within the chamber. A microcontroller unit receives the pressure signals from the pressure sensors of the handle chambers and determines the status of at least one of the device and a user of the device based on the pressure signals. A stabilization mechanism is driven by the microcontroller so as to stabilize the walker in response to the determined status of at least one of the walker and the user.

Abstract: An event-guided auto-exposure (EG-AE) device leverages the high dynamic range and low latency properties of a dynamic vision sensor to guide the exposure setting of an active pixel sensor. The method utilizes physical connections of images and events to estimate the irradiance variations, which are fed into the EG-AE device for calculating the exposure setting. The method shows outstanding performance in tackling highly challenging lighting conditions and benefits multiple downstream applications, including visual tag detection and feature matching.

Abstract: Disclosed are methods of navigating a robot in a crowd involving mapping a local crowd and applying a clustering algorithm that reconstructs a crowd-flow map capturing movement patterns of pedestrians in the local crowd; using a flow-matching metric, localizing the robot in the crowd-flow map; and following a movement pattern in the crowd-flow map using a hierarchical crowd-driven planning on a long time horizon and a short time horizon.

Abstract: A speech recognition method and related products are provided. The method includes acquiring a to-be-recognized speech and a configured hot word library; determining, based on the to-be-recognized speech and the hot word library, an audio-related feature used at a current decoding time instant; determining, based on the audio-related feature, a hot word-related feature used at the current decoding time instant from the hot word library; and determining, based on the audio-related feature and the hot word-related feature, a recognition result of the to-be-recognized speech at the current decoding time instant.

Abstract: An electromechanical sensor and a method of sensing an object or a tactile input using the sensor. The sensor includes: a base provided with a magnetic sensor arranged to detect a change in magnetic flux at the position of the magnetic sensor; a flexible film adjacent to the magnetic sensor; and a magnetic element provided on the flexible film; wherein the magnetic element is arranged to move relative to the magnetic sensor when the flexible film is reversibly deformed by an external force applied to the flexible film.

Abstract: A method and a system for end-to-end modeling are provided. The method includes: determining a topological structure of a target-based end-to-end model, where the topological structure includes an input layer, an encoding layer, an code enhancement layer, a filtering layer, a decoding layer and an output layer; the code enhancement layer adds information of a target unit to a feature sequence outputted by the encoding layer, the filtering layer filters a feature sequence added with the information of the target unit; collecting multiple pieces of training data; and training parameters of the target-based end-to-end model by using the multiple pieces of the training data.

Abstract: The present disclosure provides a semiconductor structure and a method of forming the same. The semiconductor structure includes: a substrate doped with a first ion, a deep trench structure disposed in the substrate, a barrier doped region disposed on a top of the substrate and the deep trench structure, a first epitaxial layer disposed on the barrier doped region, a body region disposed in the first epitaxial layer, a source region disposed in the body region, a gate structure disposed in the first epitaxial layer, and a collector region disposed at a bottom of the substrate. By means of the semiconductor structure, performance of an insulated gate bipolar transistor can be improved.

Abstract: An IGBT device and a method for manufacturing it, the device includes a super junction structure that has several N-type pillars and P-type pillars arranged alternately; a cell unit that is located in an N-type epitaxial layer, and the N-type epitaxial layer is located above the N-type substrate; each cell unit includes a trench gate, a P-type body region, and a source region; an N-type carrier injection layer, the N-type carrier injection layer is located in the N-type epitaxial layer, and the N-type carrier injection layer is spaced apart from the N-type substrate by the N-type epitaxial layer; the bottom of the P-type body region is located in the N-type carrier injection layer; and a collector region that is located at the bottom of the N-type substrate.

Abstract: A method for manufacturing an IGBT device includes: forming a source of the IGBT device in a substrate, wherein the substrate is an MCZ substrate; performing annealing processing on the substrate, wherein a layer of oxide is formed on the surface of the source during an annealing process; forming an interlayer dielectric layer on the substrate, wherein the interlayer dielectric layer is comprised of a silicon nitride layer, a first type oxide layer, and a second type oxide layer, and a material used to form the first type oxide layer is different from a material used to form the second type oxide layer; and performing nitrogen annealing processing on the substrate.

Abstract: A smart walking assistance device with a walker frame having generally vertical sides and an intersecting front. Wheels located at both ends of a bottom edge of the sides. A soft robotic sensing handle extends in a C shape along the upper edges of the sides and front. The sensing handle has multiple contiguous air filled chambers, each containing a pressure sensor for producing a pressure signal representing the pressure within the chamber. A microcontroller unit receives the pressure signals from the pressure sensors of the handle chambers and determines the status of at least one of the device and a user of the device based on the pressure signals. A stabilization mechanism is driven by the microcontroller so as to stabilize the walker in response to the determined status of at least one of the walker and the user.

Abstract: Disclosed are methods of navigating a robot in a crowd involving mapping a local crowd and applying a clustering algorithm that reconstructs a crowd-flow map capturing movement patterns of pedestrians in the local crowd; using a flow-matching metric, localizing the robot in the crowd-flow map; and following a movement pattern in the crowd-flow map using a hierarchical crowd-driven planning on a long time horizon and a short time horizon.

Abstract: A speech recognition error correction method and device, and a readable storage medium are provided. The method includes: acquiring to-be-recognized speech data and a first recognition result of the speech data, re-recognizing the speech data with reference to context information in the first recognition result to obtain a second recognition result, and determining a final recognition result based on the second recognition result. In the method, the speech data is re-recognized with reference to context information in the first recognition result, which fully considers context information in the recognition result and the application scenario of the speech data. If any error occurs in the first recognition result, the first recognition result is corrected based on the second recognition. Therefore, the accuracy of speech recognition can be improved.

Abstract: A method, an apparatus and a device for converting a whispered speech, and a readable storage medium are provided. The method is implemented based on the whispered speech converting model. The whispered speech converting model is trained in advance by using recognition results and whispered speech training acoustic features of whispered speech training data as samples and using normal speech acoustic features of normal speech data parallel to the whispered speech training data as sample labels. A whispered speech acoustic feature and a preliminary recognition result of whispered speech data are acquired, then the whispered speech acoustic feature and the preliminary recognition result are inputted into a preset whispered speech converting model to acquire a normal speech acoustic feature outputted by the model. In this way, the whispered speech can be converted to a normal speech.

Abstract: An electromechanical sensor and a method of sensing an object or a tactile input using the sensor. The sensor includes: a base provided with a magnetic sensor arranged to detect a change in magnetic flux at the position of the magnetic sensor; a flexible film adjacent to the magnetic sensor; and a magnetic element provided on the flexible film; wherein the magnetic element is arranged to move relative to the magnetic sensor when the flexible film is reversibly deformed by an external force applied to the flexible film.

Abstract: A method for manufacturing an IGBT device includes: forming a source of the IGBT device in a substrate, wherein the substrate is an MCZ substrate; performing annealing processing on the substrate, wherein a layer of oxide is formed on the surface of the source during an annealing process; forming an interlayer dielectric layer on the substrate, wherein the interlayer dielectric layer is comprised of a silicon nitride layer, a first type oxide layer, and a second type oxide layer, and a material used to form the first type oxide layer is different from a material used to form the second type oxide layer; and performing nitrogen annealing processing on the substrate.