Amazon Technologies, Inc.

Beijing Jingdong Shangke Information Technology Co., Ltd.

Jd.com American Technologies Corporation

Anhui Huami Information Technology Co., Ltd.

Hongda Mao

Wei Xiang

Ting Chen

Fei C Wang

Oguz Hasan Elibol

Yajun Zhao

Sundararajan Srinivasan

Jixiang Su

Chumeng Lyu

Itay Teller

Chu-Cheng Hsieh

Mohamed G Mahmoud

Weidong Zhang

Andreas Stolcke

Yuguang Yang

Eunjung Han

George Yu-Chien Lin

Zhenning Tan

Ruirui Li

Techniques for performing voice-based user recognition are described. When a device receives audio data corresponding to a spoken user input, the device may generate spoken user input embedding data representing speech characteristics of the spoken user input. The device may also identify user embedding data representing speech characteristics of a user known to the device. A machine learning (ML) model may process the spoken user input embedding data to generate reduced spoken user input embedding data including a reduced number of dimensions, where the reduced number of dimensions are functions of the higher number of dimensions of the spoken user input embedding data. Moreover, the ML model may process the user embedding data to generate reduced user embedding data including a reduced number of dimensions, where the reduced number of dimensions are functions of the higher number of dimensions of the user embedding data and are tuned to characteristics of users of the device (as opposed to a representation of all possible users). A comparison of the reduced spoken user input embedding data and the reduced user embedding data may be used to determine whether the user spoke the user input.

Voice-based user recognition

System and method for simultaneous object detection and semantic segmentation. The system includes a computing device. The computing device has a processor and a non-volatile memory storing computer executable code. The computer executable code, when executed at the processor, is configured to: receive an image of a scene; process the image using a neural network backbone to obtain a feature map; process the feature map using an object detection module to obtain object detection result of the image; and process the feature map using a semantic segmentation module to obtain semantic segmentation result of the image. The object detection module and the semantic segmentation module are trained using a same loss function comprising an object detection component and a semantic segmentation component.

System and method for real-time, simultaneous object detection and semantic segmentation

A first neural network model of a user device processes audio data to extract audio embeddings that represent vocal characteristics of a user of an utterance represented in the audio data. The audio embeddings may then be hashed to remove characteristics specific to the user while still maintaining a unique set of characteristics. The hashed embeddings may be sent to a remote system, which may use them to identify the user.

Speech processing using embedding data

System and method for semantic segmentation. The system includes a computing device. The computing device has a processor and a non-volatile memory storing computer executable code. The computer executable code, when executed at the processor, is configured to: receive an image of a scene; process the image by an encoder to form an encoder feature map; process the encoder feature map by a pyramid pooling module (PPM) to form an PPM feature map; and process the PPM feature map by a decoder to form a segmentation feature map.

ThunderNet: a turbo unified network for real-time semantic segmentation

This disclosure provides wearable-device based user-interested information determination methods, apparatuses and wearable devices. The method includes: receiving, by an electrocardiography (ECG) sensor associated with the wearable device, an ECG signal of a user, determining a feature set for the ECG signal, in which the feature set includes time-domain feature data of the ECG signal and frequency-domain feature data of the ECG signal, and determining the user-interested information based on similarity between the feature set and reference feature sets indicative of the user-interested information, in which the user-interested information includes health information associated with a disease. The wearable device includes an ECG sensor configured to receive an ECG signal and an FPGA system. The FPGA system includes modules for determine user-interested information based on the ECG signal. The apparatus includes a processor and a memory coupled to the processor. The memory is configured to store instructions to implement the method.

Determining user-interested information based on wearable device

Growing community of inventors

Fremont, CA, United States of America

Hongda Mao