This is the script used for Word Burst Rescoring and Unique Penalization reported in the work: "Features for Search and Understanding ofNoisy Conversational Speech"
This is a python based tool. In order for it to work, you need to have all the binarys used in IARPA BABEL program.
The input data is in the form of input.net, and the usage of the python script please check the shell script. This is not intended to be a full end-to-end script, more of a demostration of our algorithm.