Keyphrases
Speaker Identification
100%
Adaptive Denoising
100%
Restricted Boltzmann Machine
100%
Bottleneck Features
100%
Deep Learning Classifier
100%
Denoising
50%
Speaker Dependent
50%
Test Utterance
50%
Deep Autoencoder
50%
Two-layer
25%
Result-oriented
25%
Utterance
25%
Performance Gain
25%
Frame-based
25%
Output Layer
25%
Backpropagation
25%
Posterior Probability
25%
Extractor
25%
Deep Learning
25%
Low Dimension
25%
I-vector
25%
Autoencoder
25%
Low SNR
25%
Noise-robust
25%
Softmax
25%
Contrastive Divergence
25%
Robust Speaker Identification
25%
Bottleneck Layer
25%
Dependent Features
25%
Computer Science
de-noising
100%
Speaker Identification
100%
Restricted Boltzmann Machine
100%
Deep Autoencoder
50%
Deep Learning Method
25%
Performance Gain
25%
Autoencoder
25%
Experimental Result
25%
Posterior Probability
25%