Here's the high contrast image with the blocks blacked out solid.
http://i.imgur.com/dhOT7zP.jpg
One thing is apparent, it is in fixed length sections, sections demarked by double-width blocks going the full frequency range from 9Khz to 16Khz. I'm wondering if we have a complete signal though, as the first and last sections seem to be partial.
I did not block out the long thin elements (other than the very first, as it is half-length and the others appear full length) as they often appear to overlap slightly with the larger sections, so I am assuming they are co-incidental/background noise. They are easy to identify though, if they are important they still stand out well as elements of the signal. - toward the end there are some strong long thin lines equally spaced, which seems suspiciously deliberate, but I didn't block them out.
I also have stuck with the lengths /widths I could see in the signal, - there was a temptation to align what appear to be slightly different length/widths blocks, but I didn't do it in the end as I do not want to add any of my own interpretation into the data.
There appear to be 4 types of blocks in total.
(The image should be read with 9khz at the bottom, 16Khz at the top. The data blocks seem to stretch from 10Khz to 16Khz, the section-end blocks are full 9Khz to 16Khz)
The four blocks types are (The Khz numbers are just indications of length from the Y-axis of the original audio spectrum, the freq. range I don't think is actually important);
Blocks from 16Khz down to ~10Khz (full length) half-width
Blocks from 16Khz down to ~13Khz (half length) half-width
Blocks from 10Khz up to ~14Khz (2/3 length) full-width
Section end blocks, 9Khz up to 16Khz (full length+ a bit more) double-width
- not including the full length 9Khz - 16Khz thin 1/4 width blocks. - but looking at it more, I am leaning toward those being important, maybe as sub-section markers, though the equally spaced ones with nothing inbetween work against this theory?
looking at it, it's mostly reminding me of a stretched-out barcode at the moment. but just a guess.