This repository implements some shell scripts which investigate the performance of DNN-HMM hybrid speech recognition systems built using HTK.
Code depends on HTK 3.5.0 (pre-release as of 12/12/2015) and University of Cambridge internal helper scripts. Since these are not publically available, the purpose of this repository is to archive the ZSH shell scripts.