It is all too common for material these days to be mastered at high average levels of anywhere between -13dB to -9dB relative to full scale square. If the music you do has dynamic variation as part of the music or song structure then you should be aware of the reason why excessive limiting is inappropriate and how it alters the musical meaning of your work in a negative way. It is a process I call dynamic inversion and I here it all too often.
Dynamic Inversion
So what is dynamic inversion and how does it occur? To answer that question we will call upon a simple contrived example that reflects what often goes on in many recordings not excluding work coming out of the best mastering houses. Consider the simple case of a recording in two parts. It might me a vocal and guitar played dynamically with lets say a mezzo-piano verse and a forte chorus. Then assume the levels in the pre-limited mix are as follows:
Verse
- guitar part: -8dB
- vocal part: -8dB
- combined level: -5dB
Chorus
- guitar part: -2dB
- vocal part: -5dB
- combined level: -0.24dB
The combined levels come from adding the levels on a power basis by converting dB to mean square, adding then back to dB again. For example in the second case this equates to,
10 * log10((10 ^ -2 / 10) + (10^-5 / 10)) = -0.2357…
where the ^ operator means to the power of.
Now lets assume we do aggressive limiting on the recording such that we raise the quietest part, the verse, as loud as we can before limiting cuts in and then leave the limiter to deal with the level of the loudest part (something that happens all to0 often in loudness maximised masters). The combined level in the quietest part is -5dB so we add 5dB to bring it up to 0dB. What happens to the levels of the individual parts.
Verse
- guitar part: -8 + 5 = -3dB
- vocal part: -8 + 5 = -3dB
- combined level: -5 + 5 = 0dB
Chorus (un-limited)
- guitar part: -2 + 5 = 3dB
- vocal part: -5 + 5 = 0dB
- combined level: -0.24 + 5 = 4.76dB
Note above for the chorus we haven’t accounted for the limiter. These figures assume we can make it louder than zero but we can’t so the limiter will take off -4.76dB to bring the combined level back to zero.
Chorus (limited)
- guitar part: 3 – 4.76 = -1.76dB
- vocal part: 0 – 4.76 = -4.76dB
- combined level: 4.76 – 4.76 = 0dB
Now here comes the interesting part. Look now at how the relativity of the level of individual parts has changed from before limiting to after limiting.
Guitar Part before limiting
- verse: -8dB
- chorus: -2dB
- mezzo-piano to forte = +6dB (increase in loudness)
Guitar Part after limiting
- verse: -3dB
- chorus: -1.76dB
- mezzo-piano to forte = +1.24dB (compressed increase in loudness)
Vocal Part before limiting
- verse: -8dB
- chorus: -5dB
- mezzo-piano to forte = +3dB (increase in loudness)
Vocal Part after limiting
- verse: -3dB
- chorus: -4.76dB
- mezzo-piano to forte = -1.76dB (compressed decrease in loudness!!!)
How about that! Because we have so aggressively limited the track and the relativity between instrument parts differ between sections, we have ended up inverting the dynamic relationship between verse and chorus for the vocal part. Instead of the chorus being louder than the verse it is now 1.76dB quieter! This is what I call dynamic inversion and it annoys the hell out of me!
Even in the absence of technical inversion (when the contribution of each instrument is exactly the same between parts, eg. if both guitar and vocal went from -8dB to -2dB between verse and chorus) it still sounds like a dynamic inversion because of perception factors, particularly for vocal parts. Why? Because through our own life experience we have learned to associate loudness with timbre. When you get shouted at it is loud and the voice has a certain hard quality to it, whereas when someone is conversing with you in a friendly manner it is quieter and warmer. The same happens in singing although in this case the chorus that is loud has the timbre of a loud voice but is no louder than the verse (in dB level), which has the timbre of a quieter voice, consequently, to the brain it seems like in the verse the vocalist is singing right next to you and then in the chorus he/she has suddenly been moved 6 metres / 18 feet further away from you. My brain finds the situation really weird and truly destroys the enjoyment of the performance for me, yet I hear it happening over and over again in all too many popular music recordings.
A Real World Example Courtesy of Youtube
Much as I hate doing this for the fact that these posts on youtube are typically an infringement to the artists intellectual property, it serves as an easy way for me to illustrate the point in what is otherwise an exceptional recording, but in my view, could have been better with less limiting. The track in question is Drawing the Line by Porcupine Tree and if you like what you hear then do the artist a favour and buy the album.
Put on decent headphones, put on your critical listening caps then start listening. The track starts with a repeated xylophone like figure and a thundering, massive sounding drum part that is both dynamic and punchy and then proceeds to Steven Wilson’s vocal part which is both intimate and large. Then at around 1:25 on the timeline the chorus cuts in with Steven Wilson singing loud and gruff but note how the vocal part has now shrunk in stature because the timbre is obviously of someone singing loudly but the part is no louder, and perhaps, in terms of peak power output, probably quieter than the level of the vocal part in the chorus. The louder they play the smaller his voice sounds shrinking into the background. Then at 2:19 on the timeline he’s rushed back and is now sitting next to you singing intimately and large again. I guess some people must like this sort of thing but I’m not one of them. I love the track and it’s an exemplary recording from a great album but I just feel as though I’ve been short changed by the limiting and wished I could have heard it prior to mastering.
Now this example is from what I’d call not so aggressive limiting and very good mastering but I still come away with a feeling of dynamic inversion. For it to trouble me for this relatively good case, I’ll leave it to your imagination as to how disappointed I feel in the really aggressive masters out there, for which there are many too many.
Next time you want to put out a great recording of your music which makes use of the loud-soft idiom spare a thought as to what over-limiting is actually doing to the musical meaning of the track and hopefully lower your mastering level expectations to give it more life, space and size.
