Skip to main content

Table 2 Details of recording conditions for impulse response measurement

From: Deep neural network-based bottleneck feature and denoising autoencoder-based dereverberation for distant-talking speaker identification

Array number

Room

Array type

RT60 (s)

(a) CENSREC-4 database for training

1

Japanese style room

Linear

0.40

2

Japanese style bath

Linear

0.60

3

Elevator hall

Linear

0.75

(b) RWCP database for testing

4

Echo room (cylinder)

Circle

0.38

5

Tatami-floored room (S)

Circle

0.47

6

Tatami-floored room (L)

Circle

0.60

7

Conference room

Circle

0.78

8

Echo room (panel)

Linear

1.30

  1. RT60 (second), reverberation time in room. S, small. L, large.