ETRI-Knowledge Sharing Plaform

Self-Improving Acoustic Model Compensation Method Artificial Intelligence System Support Vectormachine(SVM) Log Likelihood Ratio(LLR) Isolated Word Recognition Confidence Measure Word-Level Within-Class Covariance Normalization Truncated Singular Value Decomposition Barrier-Free Universal Background Model Embedded Devices Frame-Level Low Correlation Low-Resource Nonlinear Manifold Objective Function Parameter Reduction Automatic Translation Distance Method Model Parameter Non-Native Speech Position Information Similar Words Training Data Vector Based Distance Algorithm Equal Error Rate Reduction Technique Channel Effect Correct Position Computing Technology Factorization Approach Robust Speech Recognition Speech Recognition System System Performance Total Variability Cosine Distance Channel Compensation Deep Neural Network(DNN) I-Vector Linear Discriminant Analysis(LDA) Matrix Factorization Manifold Structure Phone Recognition Word Similarity Best Performance Distance Measure Frequency Ratio Low-Rank Matrix Non-Similar Acoustic Features Euclidean Norm Knowledge-Based Target Matrix Computational Complexity Cognitive Computing Language Barrier Manifold Regularized Relative Reduction