This isn't going to be any help, but I thought I may as well point out that I also tried to get DNS working with IVM, but eventually gave up.
This was a very long time ago, but I vaguely remember doing some diagnostic testing and discovering it had something to do with IVM giving the wrong extensions or file formats to the audio recordings. When it did this, DNS was unable to process the file for speech recognition and would error out with something like 'Invalid file format' (error not visible). I discovered that if I manually corrected the file type (I think I may have used some audio conversion utility), and then got DNS to process the file, it worked well ...as well as a speech recognition engine usually works, which isn't too great for phone recordings. I could never get the process to be automatic though.