I'm curious what are you making?
A true, accurate, high quality (higher than the real HW) and recording professional tested DolbyA decoder.
The actual interface with the industry and certain gov't archival organizations is being done by a recording pro who I am not disclosing yet (the pros/industry people who are talking to him obviously know who he is.)
I am necessarily being careful because I have a not-so-good history communicating with people who might disagree or misunderstand, so I am mostly just playing nerd and had previously offered early (and not full quality) demo versions to the consumers. (I don't respond well to my project being called snake-oil, even though I understand why someone might all it so.)
A DolbyA decoder is not defined by equations at all, and is defined by two versions of legacy hardware (one is a diode based gain control, and the other is FET.) The specific configuration is impossible to directly do in software, so it is the job of the programmer/DSP person to properly emulate that relatively ancient design. In the past, there had been failed (or relatively more poor) designs, while even the officially un-relased versions of this decoder are pretty good, and getting better.
If only there was a better specification, it would be easier to design in the more practical (for DSP SW) feedforward design. However, I have been successful at unfolding the design and achiving nearly identical characteristics. IT IS A DAMNED TRICKY thing to correctly do.
There are some real enhancement (WRT the HW) in my design, including improved intermod and of course totally mitigated mixing with the sampling (which isn't really a HW artifact, but is a real problem for all FAST gain control software.) FAST gain control can be very tricky -- but typical SW agc tends to work around the speed issue so as to avoid the mixing with the sampling (aliasing is part of the problem.) My SW has to meet a speed spec, so my SW really has to work with 1 & 2 msec attack times (and aprx 35msec/70msec release times) and not produce intermod components mixing with the sample rate (or wrapping around zero.) The decoder is very successful at doing the high speed attack/decays without ANY audible distortion at the level that even pros can detect. Spectoraphs even show less distortion than a real DolbyA unit.
So -- that is my VERY CHALLENGING, and progressively more and more succesfully implemented project.
Perhaps the biggest bug-a-boo lately is the ability of the decoder to sometimes succesfully noise reduce material that hasn't been DolbyA encoded -- which can sometimes be confusing when I might not know which consumer recording might or might not have DolbyA encoding without being decoded!!! Happily for me, there is enough consumer material that is DolbyA encoded (some of the historical -- crunchy digital sound -- is DolbyA encoding.)
Working hard, and it is actually pretty amazing to listen to recordings that haven't been heard as nicely since the DolbyA tape was created. A DolbyA unit cannot do as well (the decoder is actually clearer, and reproduces vocal choruses with more detail, but not any frequency emphasis!!!)
John