This code and date here reflect the work described in the ACL'2018 paper on word error rate estimation e-wer.
It also contains a results for predicted WER using the glass-box features (using speech decoder features) and back box features (exeluding speech decoder features).
This data and the reported results are described in this paper:
@InProceedings{,
author={Ali, Ahmed and Renals, Steve},
title={Word Error Rate Estimation for Speech Recognition: e-WER},
booktitle={ACL},
year={2018},
}