From: Deep neural network-based bottleneck feature and denoising autoencoder-based dereverberation for distant-talking speaker identification
Method
Number of units
CENSREC-4 database
Ave.
in each layer
0.40
0.60
0.75
BF-DNN
512
84.90
78.95
83.15
82.33
1,024
88.40
83.90
88.10
86.80
2048
87.45
83.25
87.05
85.92