Just like many verbalised concepts and adjectives that attempt to describe the qualities of a mix, ‘creamy-smooth’ is an elusive and somewhat intangible term, and indeed, the non-producing folk will probably struggle to describe what it is exactly that makes a song sound pleasant and fluffy to their ears. With that said, they will likely all agree it is a desirable quality of music, and by extension one that a producer / mixer / mastering engineer should know how to achieve. (Of course, as always, exceptions apply: some genres of music deliberately use ‘excessive’, ‘unpleasant’ harshness as an assertive stylistic device.)
A great deal of the electronic music that comes my way to master is also full of character and gets its point across by using ample amounts of distortion— however, this doesn’t automatically mean songs of this nature cannot or should not strive to sound ‘smooth’, or (perhaps more aptly in the context of characterful tunes) ’aggressive in a controlled way.’
A common notion is that smoothness is directly related to airiness, and therefore achieved most easily by applying static EQ boosts in the very top-end of the frequency spectrum. Indeed, a mix that is already perfect and only lacks a sprinkle of sparkle or airiness, can be rendered silkier by using an EQ filter like this one (to name but one great in-the-box-tool, Maag EQ’s famous ‘Air Band’ knob can do wonders for the top end). An extension of this notion is the ‘smiley-curve’, which basically means that the song’s frequency response is prominent in the bass and in the top end, effectively dipping the mids slightly. And again: this does create the effect of a warm, creamy mixdown, but only if the song had been mixed in a controlled way to begin with.
In reality, and especially when it concerns electronic music that gets played at very high SPLs over loud systems in clubs and at festivals, it is rather the way the mids (and high-mids especially) that most contribute to whether a song is perceived as smooth, or if you instead walk home from the club with ringing in the ears.
Our ears are most sensitive to the high-mid region of, say, 2kHz-5kHz (you can refer to the so-called Fletcher-Munson curves if you’d like further information on this). If we manage to control the mix in this frequency region, we have essentially won the battle, in my opinion.
Even dirty, characterful tracks can sound dirty in the right way, or dirty in a wrong, haphazard kind of way. So what do I mean by ‘controlling the mix’ in this frequency spectrum? I’d advice focusing on wild dynamics and transient spikes in this frequency area. Overcooking the attacks of drum hits using transient designers for example can create such spikes. Another contributor to harshness, less transient and more sustained, can be the product of excessive use of distortion on mid-bass, for example, or loud cymbals.
Because an EQ will alter the balance of the song statically, the way to control the harshness is better done dynamically:
—Using a multiband compressor (zero in on the 2-5k area mentioned above as a starting point to tweak from);
—using a de-esser (which under the hood is one of the two above processors. One absolute killer of a plugin for this use is the Supresser by Sony Oxford).
Tip: if your dynamic processor has a dry/wet functionality, I find it useful to slightly overdo the transient- or dynamic-supression on the wet signal using short attack times, and then dial back in the dry part of the signal, finding the correct balance (as further discussed below).
Knowing the tools to tame harshness is the easy part. The most important bit is the assessment of how much you should be compressing the offending tracks or stems (or the whole mix on the master bus for that matter). This comes down to both trial and error, experience, referencing other songs, playing back on as many systems as possible, but most importantly playing back at different volumes / SPLs. If you refer to the Fletcher-Munson curves linked above, you will learn that not only do our ears perceive different frequencies with the same nominal loudness as differently loud, but the way we subjectively perceive loudness of various frequencies also changes depending on how loudly the music is played back.
What this means in practice, when it comes to controlling the mids of a mix, is that if we only monitored loud, we would end up smoothing out the mix too much— to the point that when the volume is lowered back to average listening SPL, the material would sound muffled. Finding the balance between ‘enough edginess at lower volume’ and ‘let’s not kill the ears of the people on the dancefloor’ is a moving target and a fine art. But knowing the rules of the game can now hopefully empower you to make right decision. :)