Getting something reliable out of the Piezo sounder signal was the most problemtic aspect of this project. I tried lots of things, measuring distance between peaks, magnitude and frequency at different distances - in the end, the most reliable was to average thousands of readings (I used 4000 consecutive readings); the higher the average, the further away the tag is.
I then calibrated readings against actual distances and worked out a basic formula (a sort of inverse square) to convert reading to distance. I would dearly like to find the feed that the device gets prior to piezo sounder - I think it would be more accurate and much faster.