Prepare files for ctc_decoder

Data

Text to generate the LM and lexicon should be in data.

Usage

cd scripts
bash decoder_files.sh

Different n-gram models can be created by changing the DATA_PT and N_GRAM parameters in this script.