Very interesting information - glad to hear from someone in the field (ohh, bad pun... sorry, i couldn't resist
ahh-hahaha... oh, man).
OK, seriously though... it's true what you say about searching for the best solution vs. an effective one, and every situation will have different constraints.
Admittedly, I have not personally made quantitative measurements of LF magnetic shielding before, I just try to apply what theory I've learned as best I can and hope that by following best practice and with attention to detail, I can avoid problems with what I build at home. It may be difficult or impractical to perform experimentation, especially since my DIY projects are mostly 1-off, and so there really isn't a good opportunity to make major changes once the thing is built (only tweaking, unless the problem is bad enough to warrant starting from scratch!). Finding a balance between overkill vs. 'sufficient' / cost effective design measures is definitely difficult for the hobbyist. I tend to err on the side of overkill, so long as it doesn't cost a lot.
At work, I try to apply best practice design where possible, but there are other people who pay attention to EMI for me, and I generally only have time to focus my attention on the circuit functionality of interest anyway. We have the luxury of several prototype stages to work with, plus fully equipped facilities and budget for all the toys and tweaks you could ask for.
To anyone interested in really learning the nitty-gritty of circuit grounding and shielding, I highly recommend Henry Ott's "Noise Reduction Techniques in Electronic Systems". It's been over a year since I last looked at this book, and I'm once again impressed at the quantity of good information packed in here... I've got the 2nd Ed. open in front of me, looking at an interesting graph (page 185 for those who have the book). Since it's copyright, I won't post a scan, but in essence, the graph is a comparison of near-field (0.1inch separation) shielding effectiveness (magnetic attenuation only) of various materials at different frequencies. At 1kHz, copper and aluminum sheet 40mil thick show about 5 and 3dB attenuation, respectively, while 40mil steel shows 15 or 16dB, and 30mil mu-metal around 18dB. At 100kHz, steel is only slightly better than copper. This data doesn't show to the difference at 60Hz, but one can extrapolate... at 50/60Hz, the difference is bound to be larger than 10dB. But it's still good to know you can get 15 or 20dB out of 1/4" Al if that's what you happen to have on hand (I know a lot of hobbyists use Al plate stock for chassis construction). Ott goes on to say later in the chapter that "Actual shielding effectiveness obtained in practice is usually determined by the leakage at seams and joints, not by the shielding effectiveness of the material itself". If using conductive material, seams should have as much overlap as possible, and should support good conduction across the joint. Therefore, bare aluminum will do better than anodized.
I guess the point I was trying to make earlier was that when it comes to dealing with low-frequency magnetic fields in an audio system, I would first turn to some basic design principles before I would consider shielding materials. Shielding would be icing on the cake, but maybe not so necessary if the other pieces are done right. Also consider if you need to shield the entire power trafo, or just install a small shield around a particularly sensitive circuit.