Trying to help...
According to the information I've been able to gather in the last three minutes:
It appears that:
Sound transmission loss (TL) is a laboratory measure of a barrier's ability to reduce the flow of sound energy through it. It is the ratio of the sound energy striking one surface of the barrier to the energy radiated from the opposite surface, and it ignores diffracted sound that bends around the top or ends of a free-standing barrier. The primary determinants of TL are mass (lb/sq. ft.) and leaks (air paths through the barrier). Essentially any barrier material with a mass of at least 4 pounds per square foot that is also free of leaks will function well as an outdoor, free-standing noise barrier. TL is measured by ASTM E-90 which utilizes a sealed opening fitted with the test panel. Its relevance to an outdoor, free-standing barrier is minor.
Sound transmission class (STC) is a rating number derived from TL readings at 16 different frequencies, in accordance with ASTM E-90 and ASTM E-413. It is a laboratory measurement in a sealed opening taken in such a way that there is no diffracted sound reaching the receiver. Plywall has been tested per ASTM E-90 and E-413, resulting in a Sound Transmission Class of 38.
-- From http://www.frtw.com/products/noise/pw_tech_acoustical.html
After reading this, it appears empirical, since lab work is done.
Boring stuff after this point.
Hypothetically, if you gave me some untested (eg random or theoretical) TL readings for 16 different frequencies, I could pump out a calculation based on that information. This STC would definitely be theoretical. If I took an actual measurement of TL and then derived the STC via equations and actual data, then the TL would be empirical but the STC would be theoretical.
But, if the website is correct: (STC)"...laboratory measurement in a sealed opening..." Then, since they are doing measurements with instruments, then it is empirical.
I think I almost confused myself. I think the mystery is in the definitions of theoretical and empirical.
Another way to look at it:
A) A circle with radius x.
B) Using radius x, calculate the circumference of the circle. This is not empirical since no measurements were done.
C) Using tools and a whole lot of patience, measure the circumference of the circle. Empirical.
B and C could be the same result. If the input is right (A), I don't see a need to measure (C), when you can just use the equation(B).
So I guess you asked a loaded question. The relationship between TL and STC is not fixed at either theoretical or empirical.
(Now of course, since I have no experience with these two things, I could be completely wrong. Someone let me know if this doesn't jive.)