Originally Posted by Ed Mullen
You can double check with Chris K - but I believe Audyssey does the level match and time alignment from the primary LP on the first measurement.
What measurement system are you using to conclude Audyssey sets the subwoofer channel 4-5 dB lower than the speaker channels?
I have noticed that Audyssey sometimes doesn't hit exactly 75 dB on the speaker channels, so you have to check the speaker channels and the sub channel to determine the relative delta (if it exists) between them. If you are using a C-weighted SPL meter, add ~3 dB to the meter reading to get the actual unweighted SPL from the sub channel.
The Disney WOW DVD uses the LFE channel for the subwoofer rumble tone. Presumably that tone is encoded 10 dB lower than the speaker channel tones, but I can't speak to that as factual.
The Disney WOW disc tones are all recorded at -20 dbSF (as confirmed by Roger Dressler - see the below link), so there is indeed a 10dB delta between the sats and LFE. In this case, the disc fails to account for the 10dB goose to the LFE by the AVR (WOW is actually 10dB too hot). After letting Audyssey do its thing, I double checked all of the sats to make sure they were at the same level (using a Galaxy 140 SPL meter at the MLP), and they were all pretty much spot-on. The LFE, however, came in only 5dB higher (allowing for the 3dB shortfall) instead of the expected 10dB, so I had to raise the trim by about 5dB just to get flat with the sats. At the end of the day, I have the gain on both subs set to -14 (the best I could do before my AVR trim was buried at -12), Audyssey set the AVR trim to -10.5, I had to raise this to -5.5 for a flat calibration, and further goosed it to -2.5... In total the final trim is +8dB above where Audyssey put it, but the net result is that I am running +3dB hot relative to a flat calibration. Both MultiEQ and XT32 yielded similar results.
"Official" Audyssey thread (FAQ in post #51779)
Even checking against SD-DVE from back in the day (which has a well-documented 10dB discrepancy) yields the same results. Of course, that is only in my room, and every room is different. But, given the number of similar reports (concerning Audyssey setting LFE too cold), the few reports of Audyssey's final LFE trim setting being correct with no tweaking needed (I even saw one where the guy had to lower
the trim to taste...?!?) seem to be the exception...
If anybody can see a problem that I missed, it would be you. Here are my before/after FR graphs: