Skip to main content
. Author manuscript; available in PMC: 2016 Mar 9.
Published in final edited form as: IEEE/ACM Trans Audio Speech Lang Process. 2015 Jan 14;23(1):92–101. doi: 10.1109/TASLP.2014.2372314

Table III.

WER on the development set (si_dt_05) of the CHiME-2 corpus when the same data is used to train the mask estimator and the acoustic model

α si_dt_05
−6 dB −3 dB 0 dB 3 dB 6 dB 9 dB Average
0.0 42.7 33.9 28.9 25.2 21.7 19.2 28.6
0.5 38.4 30.5 26.5 22.9 20.8 18.9 26.3