My mids/highs are 5-10db overcompressed?

[audiovisceral]

I just purchased a copy of Har-bal and have so far found it to be a wonderful tool so far for identifying rogue frequencies. Additionally, it's given me a whole new perspective on numerous aspects of my mixing/mastering technique. In doing so, I've discovered an issue I'm hoping someone can help with.

Har-bal has shown me that many of my mixes/masters are coming out with, at times, only a 5-10 dB difference between the peak and average power in the 500-8k range. Pro songs at a similar RMS, by comparison, tend to demonstrate one and a half to double this range mine show.


To demonstrate:
Here is a screenshot (with professional reference) and short audio clip of a typical master:
http://www.audiovisceral.com/avmaster.gif
http://www.audiovisceral.com/avmaster.wav

Here is a raw mixdown of the same song (has a bit less track compression on the drums):
http://www.audiovisceral.com/avraw.gif
http://www.audiovisceral.com/avraw.wav


My mastering chain is:
1) BBE Exciter
2) LinEQ/Waves Q (eq based on Har-bal guidance) - not yet performed in track above
3) TC NEss softnee compressor - 2 or 3 dB control compression to catch spikes
4) LinMB - Too Much Limiting with attack 90, release 100 on the low freq.
5) L3 - 5-9 dB of reduction

In general, I am happy with the overall RMS curves of my mixes/masters and thus their general sound. However, as a result of this issue, I have noticed that they sound overcompressed through the mids/highs, and I wish I could easily get my mid/high peak power up to where it should be.


What I've tried:
I have tried applying in my mastering chain LinMB upward expansion on the squashed range. However, I have not had luck accomplishing the 5-10dB or so of expansion I need to get it to a normal dynamic range. Additionally, doing this seriously alters the sound of the my songs.

I have also tried to correct the issue at track/mix level by varying (and even eliminating) individual track eq/compression. Another option I've considering is that I may be able to apply frequency-specific expansion (low attack/release to boost the transients) on virtually all the individual tracks. Both approaches have changed the sound of my music, but not significantly increased the separation I want to see between those two Har-Bal lines ... I feel at a loss.


My questions:
1) How should I interpret the squash between peak and average power in my mixes/masters?
2) Given the audio/screenshots above, what stage is this a problem to be fixed during?
3) If remixing is the key, which instruments might be causing the problem, and what sort of processing should I be aiming for?
4) If remastering is key, how can I best do so? Would faster attack/release or other different compressor settings be the solution?


I could really use the help with this, and it seems that everyone here is pretty supportive of DIY mastering.


Mike

[HarBal]

Are you comparing like with like? If your track isn't musically dynamic (ie. the music itself and the way it is being played) and the pro one is then that will attribute to some of the difference. Are your tracks from real instruments or samples? If samples, then you may find that they've already been heavily compressed.

My suggestion would be to try taking out "all" compression from your chain and just applying modest limiting at the end. If you still aren't getting the dynamics you want after that you may have to re-record it.

Regards,


Paavo.