PACKAGES
  1. Computing feature files from speech signals: Mel-frequency cepstra, Mel-frequency logspectra, powerspectra
  2. Learning a Gaussian mixture density from a set of feature vectors
  3. Classifier based speech segmenter
These packages are primarily for my own use - I often need them, have too many versions and copies, and forget where they are located. You may use these freely for any purpose, but I am not liable for any damages you may accrue or inflict as a result. The packages work on linux machines. I can say nothing, and know nothing, about windows machines. There is a brief readme file with every package that might be instructive. Please note that due to my extremely busy work schedule, I will not be able to provide technical support for these packages. You can find answers to most of your questions, however, by simply studying the code.

If you wish to acknowledge any of these packages in your publications, website etc., you can give the reference as "[Packagename], CMUSphinx websites for robust speech recognition, Carnegie Mellon University"