You can pretty reliably depend on Olivia Rodrigo to pull some nice arrangement stunt in every song she does, and this one’s no exception. Here it’s the song’s hook line (first heard at 1:06) that’s the highlight. Not only is there a dramatic backing tacit underneath it (which is always a reliable way to get any line to pop out and demand the listener’s attention), but Rodrigo also dramatically switches to falsetto at a moment where, following 16 inexorable bars of build-up in the vocal performance intensity, you’d expect her to be going at full belt. And it’s not just a vocal-technique cop-out either, because it’s clear from the middle section that she’s perfectly capable of hitting those high notes in chest voice if she wants.

But the thing that most disproportionately caught my ear in this production was a tiny detail: the distractingly clumsy scansion halfway through the first verse, where she accents the on-the-beat definite article in the line “All the girls at this party are so cool”. At my most charitable, I might suggest that the awkward word-setting is to highlight the protagonist’s lack of cool by comparison with those other revellers, but that feels like a bit of a stretch when it’s coming from a certified pop fashionista… Now, don’t get me wrong, there’s a long history of singers syncopating natural word stresses so that they deliberately don’t land on the beat, and it’s a technique that can be really effective. However, the secret to making this work is for the singer to maintain the natural stress of the lyrics despite the syncopation – rather than unnaturally overlaying the metre’s stress pattern onto the words, as in this case.

All of which brings me to the subject of level automation. Although the most basic purpose of riding the lead-vocal fader at mixdown is simply to maintain a subjectively consistent level for the singer in the balance, there are plenty of situations where it makes sense to sacrifice absolute consistency for the sake of a more musically appropriate result. You might turn up the most expressive moments of the vocal performance to more emotionally engage the listener. You might turn up consonants and vowel transitions to improve lyric intelligibility. You might turn down quieter syllables to enhance the dramatic impact of neighbouring high-intensity notes. And, in this specific situation, you might also use your automation to reshape an awkward lyrical stress pattern into a more musically appropriate contour.

So, for example, here’s a short section of this song as it appears in the release version, complete with the “All the girls at this party are so cool” line. play_arrow | get_app And now, by virtue of a little behind the scenes editing on my part, here’s that same section, but where I’ve changed the stress pattern to lay more emphasis on the word “all” and less on the word “the”. play_arrow | get_app Personally, I think this makes the line flow better – more like an intentional syncopation than a scansion error. Whether you agree with me or not in this case, though, my general point still stands: level automation has the power to alter the musicality of a singer’s phrasing, not just their mix balance.