We describe the latest improvements to the IBM English conversational telephone speech recognition system. Some of the techniques that were found beneficial are: maxout networks with annealed dropout rates; networks with a very large number of outputs trained on 2000 hours of data; joint modeling of partially unfolded recurrent neural networks and convolutional nets by combining the bottleneck and output layers and retraining the resulting model; and lastly, sophisticated language model rescoring with exponential and neural network LMs. These techniques result in an 8.0% word error rate on the Switchboard part of the Hub5-2000 evaluation test set which is 23% relative better than our previous best published result.
Cite as: Saon, G., Kuo, H.-K.J., Rennie, S., Picheny, M. (2015) The IBM 2015 English conversational telephone speech recognition system. Proc. Interspeech 2015, 3140-3144, doi: 10.21437/Interspeech.2015-632
@inproceedings{saon15_interspeech,
author={George Saon and Hong-Kwang J. Kuo and Steven Rennie and Michael Picheny},
title={{The IBM 2015 English conversational telephone speech recognition system}},
year=2015,
booktitle={Proc. Interspeech 2015},
pages={3140--3144},
doi={10.21437/Interspeech.2015-632}
}