Skip to main content

Table 1 Scaling factor α for each type of WS-HEQ

From: Intra-frame cepstral sub-band weighting and histogram equalization for noise-robust speech recognition

Structure I

Structure II

Method

Optimalα

Method

Optimalα

WS-HEQ I ( 1 )

0.6

WS-HEQ II ( 1 )

0.6

WS-HEQ I ( 2 )

0.6

WS-HEQ II ( 2 )

0.6

WS-HEQ I ( 3 )

0.5

WS-HEQ II ( 3 )

0.7

WS-HEQ I ( 4 )

0.7

WS-HEQ II ( 4 )

0.6

  1. It gives the optimal recognition accuracy for each WS-HEQ variant in the development set as to the Aurora-2 database.