I've been reading this:
GUIDE: How To Properly Encode Dolby Digital Audio (AC3)
http://forum.doom9.org/showthread.ph...threadid=56020
And I wondered, since LAeq is a measure of loudness, and ReplayGain is also a measure of loudness, perhaps RelayGain could be a more exact approach to setting dialnorm than using just the RMS. I mean, Replaygain takes in account spectrum "weigthing", the statistical distribution of the signal etc. while RMS is much simpler.
Also, in the
www.replaygain.org site there's a link to a command line app "Wavegain" that could make it easy to use the replay gain level in automatic tools like DIKO.
Remember that the number ReplayGain puts out is how many dB of amplification/attenuation you need to reach a preset level (89dB in a scale where a -20dB RMS pink noise is 83dB), so you have to do some arithmetic with the RG to use it as an approximation for dialnorm. (If I did not err, Dialnorm=-RG-14)
I did a simple test with a rock concert avi, the RMS=-18dB (average RMS calculated by CoolEdit), and the RG calculation gave -10dB. That's quite a difference, and I don't know which would be the better value to use.
Perhaps we could do more tests with different programs, like normal movies with dialogs, music and effects, documentaries, TV captures, etc.