I haven't tried that particular product but I've used similar packages and even written some primitive software in that problem space for my own education.
Basically the technique used by all of these programs is to use LaPlace transforms to convert the audio from time domain to frequency domain - thus producing a "waterfall" of the frequencies in any given sample. It gets pretty complex because you have to trade off between reasonable sample size and a reasonably precise frequency slice.
Once you have the waterfall of frequencies then it is fairly simple coding to go through and examine the intervals between the strongest notes and come up with the most reasonable "guesses" at the chords - often times using "hints" such as certain chord combinations being more likely and so on.
Unfortunately, real world music tends to have an awful lot going on at any one time - this is especially true of modern rock and pop music where you have all kinds of harmonics flying around from the overdriven instruments. Then, add voice for some truly complex waterfalls!
I've found that every program I've tried that attempts chord guessing does somewhere between poor and middling. I finally decided that a program like "Transcribe!" is far more useful as you can get it to display the waterfall over a piano keyboard. It can also be instructed to attempt to guess chords but it's not any better at that than the other programs are. However, I find that I can usually figure out the chords from the waterfall better than a program can - our brains still have far more processing power than hour home computers.
Also, with a program like Transcribe! you can view the audio waveform and manually select samples so it's easy to force it to produce a waterfall for a very specific sample.
Someday someone is sure to solve the complexity issues and bring us a better chord guesser - maybe this is it - I'd love for that to be the case!
John
John