Preprocessing Data

If you choose to store preprocessed speech files, then you will have to do the preprocessing with JANUS or with the featmaker program. Both can use the same feature description file. Make sure that you have prepared a working feature description file, that works for your raw data. The last line in the feature description file should be one that writes the preprocessed data into a file. Write a script (shell script tha calls featmaker for each of your files, or Tcl script that calls the feature module of JANUS) and run it.