I have a stereo voice track I am cleaning up for a podcast and one of my main concerns for the client is returning a file with equivalent dB levels. I have seen this on previous files from the client so I thought I would ask because I do not understand why I am getting a dB boost. I'm probably missing something easy here.
 
The raw file as rec'd from the client shows a peak in Sonar at 0.0 dB. I cleaned up the audio with SoundSoap Pro and bounced to track and the audio peak is still 0.0 dB.
 
I instantiated Sonitus EQ in the effects bin and set it up with a HP and a LP as follows:
 
https://www.dropbox.com/s/knkzv2zyuetiysp/Screenshot%202014-06-23%2018.58.50.png
 
You can see my audio now peaks at 2.0 - 2.1 dB. Since I have only cut frequencies, why the 2 dB boost?
 
Thanks!
 
 

Look at your level meter. See the red dot. You have gone over. 
You should aim for just a tad under 0db. 
And myself I'd be using a compressor as that track looks like it's peaking a LOT. 
Hi pass is always a good Idea but if there was no low end content to begin with it won't change anything. 
Most mixers and pre amps have a hi pass filter so the signal was probably already processed. 
 
Once again I hate repeating myself but a program like Wave lab is easier to work with for wave editing because that's it's design purpose. Sonar is the worlds greatest multi track recorder but I use Wave lab for stereo editing. There are many other wave editors and some are even free like http://www.wavosaur.com/
Not saying you can't do it in Sonar, but without the analyzing tools your working blind. 
 
Example load the file. 
Check the peak level with the normalize tool. 
If the file has headroom say it's shows -4.5db then normalize it to -.05 db. 
If it say's 0 then it may be OK, but what about overs? 
This is where the Global Analyzer is indispensable because it will point them out for you. 
It will also check the average RMS level which to me is most important of all.
Average is a much different reading than peak. A file can have a reading of 0 db peak but it's average RMS level will be anywhere from 8 to 20 db. 8 db being about as loud as it gets, 20db might be a classical piece.  
If its RMS average is around 14 - 16 db it will be in the ball park on most playback systems. 
If it still shows 18-20 average RMS it's might be too quiet and you then proceed to use the Loudness Maximizer or your favorite limiter to kick it up 2 or 4 db,, no more if you can help it. 
If the file is that uneven in volume then there's some manual editing to do first.
That's chapter 2. 
 
If the file really must be output from Sonar then a great and easy solution is the LP64 Multi band compressor on the Master buss. It works pretty flawlessly to catch overs because it is a "look ahead" plug in. But that track would need to be turned down first it's over and that is not good. You could put a compressor in the tracks bin with a super fast attack, but I like my tracks to be under control without one. You can simply turn down the volume or gain. 

Cactus Music
Look at your level meter. See the red dot. You have gone over. 
You should aim for just a tad under 0db. 

But that is the point.
 
The original file peaks at 0.0 dB
 
When I add a HP & LP, the file peaks at 2.0.
 
Why? They are only EQ cuts.
 
PS Thanks for all the advice, I really appreciate it. The above was only the beginning of my treatment. I reduced the sonitus output by 2 dB, bounced to mono (after reducing by another 3 dB to compensate), brought back into sonar and used Ozone's EQ boost, compression and a limiter (starting with the Podcast, Loudness Maximizer preset) and just now gave the client a pristine (and louder by RMS) voice track peaking at -0.3.
 
I just don't understand why the EQ HP & LP increased the peak levels by 2 dB?

OK I see what your saying. I guess the only answer would be the Sontitus EQ is buggy and is adding a little gain even when set on 0. Would be easy to test this. 

Cactus Music
I guess the only answer would be the Sontitus EQ is buggy and is adding a little gain even when set on 0.

Well that is a good enough answer for me (better than my brain is addled), and definitely something for me to closely monitor as I receive more files from the client.
 
On the next file I'll try various EQ's and see how they compare. I use the Sonitus just because it's there.
 
Thanks!

Yes me too. so I'd be interested to see what happens if we plunk it on a test tone set at a known db level. 
Only thing I could see happening is if it was in parallel it would add a doubled signal path. But in series this should not happen.
 
One would assume that if it's output is set at 0 that it would not go over that unless more than 0 was input.
 

I suspect there is a slight resonant bump in the filters so in this case the signal goes over. Normally this bump would not be an issue because the filtering would happen earlier in the signal processing when the incoming signal was not so hot. EQ is usually not at the end of the mastering chain where a signal is so high.

My thinking too. As normally I would never have any of my tracks that close to zero anyhow. But it is good to be aware that if you add that particular EQ you can expect a jump in level.

I have had this sort of thing happen before and it taught me a valuable lesson about putting in eq last in a chain if I am pushing close to zero. I think scook nailed it. It only takes the smalles resonant bump by one of the filters when you are near zero to take you over the edge.

Change the slope of the hp/lp filters and see if that changes the level back close to 0db. The steeper the slope of the filter, the more the resonant spikes.