USING A DISCRETIZED, HIGHER ORDER REPRESENTATION OF HIDDEN DYNAMIC VARIABLES FOR SPEECH RECOGNITION
Patent №
US 7,680,663
Granted
2010-03-16
Filed 2006
Owner
MICROSOFT CORPORATION
Lab
AI components
4
ml · nlp · speech · planning
Assignment
Recorded
Dataset
AIPD
2023_r1 edition
Application
11507169
A hidden dynamics value in speech is represented by a higher order, discretized dynamic model, which predicts the discretized dynamic variable that changes over time. Parameters are trained for the model. A decoder algorithm is developed for estimating the underlying phonological speech units in sequence that correspond to the observed speech signal using the higher order, discretized dynamic model.
AI classification
Speech1.00
Machine learning1.00
Natural language1.00
Planning0.65
AI hardware0.15
Vision0.08
Knowledge representation0.04
Evolutionary computation0.00
Ownership
MICROSOFT CORPORATION
assignment · 182620615
Assignors
DENG, LI
On an employer assignment, the assignors are typically the inventors.
From the same owner
EYE AND HAND TRACKING UTILIZING LENSLESS CAMERA AND MACHINE LEARNINGUS 12,693,739 · 2026SECURITY ACTION BASED ON ANOMALY DETECTION USING AI MODEL PROFILES AND USER PROFILESUS 12,694,097 · 2026MULTI-STAGE MULTI-HOP NATURAL LANGUAGE AND MODEL EXECUTION PLAN GENERATIONUS 12,694,230 · 2026MODELLING CAUSATION IN MACHINE LEARNINGUS 12,694,263 · 2026MIXTURE OF EXPERTS MODELS WITH SPARSIFIED WEIGHTSUS 12,694,265 · 2026EXTREME CLASSIFICATION PROCESSING USING GRAPHS AND NEURAL NETWORKSUS 12,694,275 · 2026ENCODING IRREGULAR SHAPES USING ANGLE-BASED CONTOUR DESCRIPTORSUS 12,694,545 · 2026ADAPTIVE TELECONFERENCING EXPERIENCES USING GENERATIVE IMAGE MODELSUS 12,694,587 · 2026