Frame Skipping With Extrapolation and Outputs On Demand Neural Network For Automatic Speech Recognition
Patent №
US 9,520,128
Granted
2016-12-13
Filed 2014
Owner
INTEL CORPORATION
Lab
—
AI components
3
ml · speech · hardware
Assignment
Recorded
Dataset
AIPD
2023_r1 edition
Application
14493434
Techniques related to implementing neural networks for speech recognition systems are discussed. Such techniques may include implementing frame skipping with approximated skip frames and/or distances on demand such that only those outputs needed by a speech decoder are provided via the neural network or approximation techniques.
AI classification
Machine learning1.00
Speech1.00
AI hardware1.00
Natural language0.18
Vision0.01
Knowledge representation0.00
Evolutionary computation0.00
Planning0.00
Ownership
INTEL CORPORATION
assignment · 366610357
Assignors
BAUER, JOSEF G., ROZEN, PIOTR, STEMMER, GEORG
On an employer assignment, the assignors are typically the inventors.
From the same owner
DYNAMIC INCLUSIVE AND NON-INCLUSIVE CACHING POLICYUS 12,693,977 · 2026RECOGNITION, REIDENTIFICATION AND SECURITY ENHANCEMENTS USING AUTONOMOUS MACHINESUS 12,694,657 · 2026DEPTH AND MOTION ESTIMATIONS IN MACHINE LEARNING ENVIRONMENTSUS 12,694,659 · 2026Apparatus, device, method and computer program for an autonomously acting machine or a computer systemUS 12,686,127 · 2026Multi-tile Memory Management for Detecting Cross Tile Access Providing Multi-Tile Inference Scaling and Providing Page MigrationUS 12,688,146 · 2026Generating Pretrained Sparse Student Model for Transfer LearningUS 12,688,422 · 2026OUT-OF-DISTRIBUTION DETECTION USING A NEURAL NETWORKUS 12,688,676 · 2026METHODS AND APPARATUS TO DETECT A REGION OF INTEREST BASED ON VARIABLE RATE SHADINGUS 12,689,741 · 2026