Thanks for that.
12:1 = 96MB is mathematically impossible and the two things he is saying are in complete conflict with each other.

Take his 12 and stick a decimal point in between the 1 and the 2 and you get for the first time a realistic figure that mathematically balances perfectly.

My bet is that the compression is at it's worst is 1.2:1 when loading lost of small files, and at it's best 1.5:1 when using longer samples.

And it just so happens 1.5:1 = 96MB which is in line with the test you just did.

lol... could it be that all this was just because AJ thought is was looking at 12 instead of 1.2

Regards.
James.