Inventor

LI JINYU

US105 patents

⚠️ This page may combine multiple inventors who share the name “LI JINYU”. Patents are grouped by organization below to help distinguish them — per-person disambiguation is on the roadmap.

LI JINYU

MICROSOFT TECHNOLOGY LICENSING LLC

Speaker-invariant training via adversarial learning

Low-footprint adaptation and personalization for a deep neural network

Multi-channel speech separation

False face representation identification

Domain adaptation in speech recognition via teacher-student learning

Advancing word-based speech recognition processing

Code-switching speech recognition with end-to-end connectionist temporal classification model

Adversarial teacher-student learning for unsupervised domain adaptation

Speaker recognition

Shared hidden layer combination for speech recognition systems

Pre-training with alignments for recurrent neural network transducer based end-to-end speech recognition

Machine learning model with depth processing units

False face representation identification

Restructuring deep neural network acoustic models

Generating and using text-to-speech data for speech recognition models

Biometric recognition

Conditional teacher-student learning for model training

Universal acoustic modeling using neural mixture models

Learning student DNN via output distribution

Attentive adversarial domain-invariant training

Attentive adversarial domain-invariant training

Small-footprint deep neural network

Variable-component deep neural network for robust speech recognition

Canonical training for highly configurable multilingual speech

Training and using a transcript generation model on a multi-speaker audio stream

Speaker adaptation for attention-based encoder-decoder

Convolutional neural network with phonetic attention for speaker verification

Sequence-to-sequence speech recognition with latency threshold

Internal language model for E2E models

Convolutional neural network with phonetic attention for speaker verification

Spoofed face detection

Training and using a transcript generation model on a multi-speaker audio stream

Machine learning model with depth processing units

Speaker adaptation for attention-based encoder-decoder

Condition-invariant feature extraction network

Condition-invariant feature extraction network

MICROSOFT CORP

Exploiting heterogeneous data in deep neural network-based speech recognition systems

Multilingual deep neural network

Managed biometric identity

Managed biometric identity

LI JINYU

VGPU: a real time GPU emulator

Adapting a compressed model for use in speech recognition

Model training for automatic speech recognition from imperfect transcription data

LEYVAND TOMMER

Biometric recognition

Visual based identitiy tracking

WEI YICHEN

Real time head pose estimation

DERNIS MITCHELL

Voice-body identity correlation

YU DONG

Confidence calibration in automatic speech recognition systems

TONG XIN

Single-pass bounding box calculation

BOE TECHNOLOGY GROUP CO LTD

Shift register and driving method therefor, gate drive circuit and display apparatus