I had a look at your article. I'm a neophyte when it comes to understanding RMS meters, so I'm not so sure I understood much. But let me ask one question. I think I read somewhere in your article that with practice one can make gain staging adjustments by visually observing the wave.  When visually looking at a wave I see the peaks. I also see the condensed energy, which is often well below the peaks. Is RMS the average of this concentration of energy below the peaks? 
 
Thanks for the article.
 
Dean

Hi Dean. Sorry for being slow in responding.  Yes many signals consist of two parts.  The transient peak at the very start of the sound followed by the rms component which follows.  The meat of the sound say.  The rms component is longer in time and often by a much longer time.  The peak component is often very fast and short lived but it does rise to a higher value also as well.  Some sounds may not have a transient peak at all such a cello note for example with a slow attack.  It would have mainly the rms component only.  So all of the waveform (once it settles down to a value that is) would be the rms part of the sound.
 
The rms VU meter type measures the rms part and is also slow in responding too.  It takes 300 mS to reach 0 dB VU on the scale so peaks slip past a VU meter and make it move very little which is actually a good thing.  The peak meters in our DAW's are fast and they can show the peak values very well.  Or show how high the peaks or transients can go which is also a good thing.  The two together work well.
 
By eye what I am meaning there is say there are a bunch of audio clips on a track.  In Studio One the gain of each clip can changed and the waveform height changes as you make these adjustments which quite frankly is better and faster too.  This means it is possible to make rms components of various waves look at a similar height.  Of course I still would run a VU over them all to confirm and fine tune.
 
I just prefer to keep all the rms parts of the signals everywhere at a consistent level in the Daw system. 

Article was also linked with Recording Hound online magazine.  Good stuff Jeff.

Thanks for agreeing to let me post your article Jeff. Good stuff for sure.

The more the merrier.  Maybe Cake can sticky the article so when someone asks about gain-staging we can tell them to read Jeff's article. :-)
 

Jeff Evans
 
I just prefer to keep all the rms parts of the signals everywhere at a consistent level in the Daw system. 

Jeff, thanks so much for taking the time to respond to me. You made understanding what RMS is very clear and I get it now.
 
I have a few more questions, if you have the time.
 
1) what would be an ideal RMS level, and should all track signals in a project be set to this recommended level to achieve what is refereed to gain staging?
 
2) If I set the RMS at the recommended level there may be some peaks that hit the ceiling. What do you recommend to do regarding taming the peaks? Compression or limiting?
 
Thanks so much Jeff! You cleared up some of my confusion and I will re read this great article on gain staging again. I'm am sure I'll get more out of it this time around.
 
Dean.
FreeEarCandy 
 
 

I like to work with K System levels which are -12, -14 and -20.  The often used industry standard is -18.  Even when I work at -14 I rarely clip with the transients.  If you are concerned -18 or -20 will work great.  You will rarely clip at those levels. 
 
With drum sounds though and very fast percussive sounds I don't use the VU meter.  It is too slow and things can easily clip.  I let go of the VU meter in these situations and use peak metering instead and just ensure that the loudest hits are only reaching -6 max.  Once you send a bunch of fast transient sounds e.g. drums to a buss you will find the VU will start move quite well over a buss.  There is enough rms information in an overall drum sound to make a VU move almost normally.  Especially if some of the drums have more open tunings and hence longer decay times.  Once you tame the dynamics with a compressor of an overall drum sound, the transients will drop down a little more and the rms part of the sound will start to come up. 
 
I work at various ref levels depending on what I have to do.  My Yamaha digital mixer is calibrated for -14 = +4 dBu at the output so that is a common level for me.  All the metering in the mixer matches the DAW for me at that level, so I call it a general workhorse level.  It seems to combine both transients and being fairly loud at the same time.  For more pristine and transient type stuff definitely drop down to -20.  But you have to recalibrate the VU meter when you do this.  (and everything in my mixer will be 6 dB lower now)  No biggie though.  Once I select a ref level I stick to it for everything. Tracks, buses and the main mix.
 
But if you do a -20 production and you want to end up with say a loudish master e.g. -10 rms mastered, then you have got to bring up the rms level of your premastered mix a whole 10 dB which is quite a lot.  Can be done though.  But if you are already at -14 then you only have to lift your overall rms by 4 dB now which is easier in mastering.  Nothing has to be pushed real hard in order to gain an overall 4 dB of rms level in your mix.  I try and think a head a little before choosing the ref level. 

Do mix engineers traditionally set different VU calibration levels for individual tracks, effect buses, group buses, and the master bus? Or should all be kept at the same level? 
 
And thanks for posting that article. Coincided with my current interest.

Thanks, I'll have to keep this in mind.

I prefer to keep the calibration levels constant in all areas. At -20 for example it is easy to track your tracks there. It is also easy to send multiple tracks to a buss but each track will be at some point a bit below unity often. The buss level can also be at -20. Multiple buses can also often be sent to the final mix and for me that ends up at -20 as well.  This keeps the headroom consistent all the way through.  Another reason is if you have hardware VU meters for example if you monitor a track it is just hitting 0 dB VU nicely.  If I monitor a buss the VU will also be just reaching 0 dB VU and when checking a whole mix the VU's will also be hitting 0 dB VU.  Nice consistency for the VU meter too in all these parts of your production.
 
Another good reason for consistent ref levels in your key production points, is for SPL levels too. If you tie your ref level to a specific monitoring SPL level, e.g. 85 dB SPL then when monitoring tracks, buses or the final mix,  you will be hearing each of these at the same SPL level. The tracks sound a little louder at the same ref level but as you are wanting to listen in a little deeper with them it is not a bad thing. Buses are starting to contain a few tracks so the more stuff in there, the slightly lower it may seem in volume. The mix has got everything so when it is producing 85 dB SPL, it all seems at a slightly lower level again in SPL's.  The mix ends up brought back so it is easy to listen to.  All that is going on here is that as you monitor areas with more music present, the SPL levels are being maintained at these points.  It all starts make sense.
 
But no harm in using different calibration levels either. The VU's planted there would need to be re-calibrated.  Not so good for the hardware VU's in my case.  (that prefer to remain on a single cal level at least) The ref level that reduces the headroom will set that headroom from there on. You could work at progressively louder rms levels though from one stage to the other. e.g. tracks at -20, buses at -14 and the final mix at -12. When you mix into something like Waves CLA for example, it is somewhere between a pre mastered mix and a fully mastered mix.  Great for sitting things up at -12 or even up to -10 dB rms.  It is quite good to mix into something like CLA for example. You could print two versions of the mix, one with and the other without CLA.
 
For an album it is good to mix each track to the same rms level. They will then be in good position for mastering.