I let this song lie a little while before posting about it, because I somewhat foolishly released it not that long before Christmas when I was about to get pretty busy and distracted – as were most potential listeners. It’s a simple, folky, somewhat stupid little song about a semi-professional Hearthstone player who plays under the alias Zetalot. (Hearthstone: a popular digital card game set in the universe of the Warcraft games that I grew up on.) I wrote the song on the plane back from a weekend in New York and had it fully recorded and mixed a month later, and the thrill of working that efficiently was a big part of what made me want to release it almost immediately (when saving it for January might have been more savvy.)
I recorded the song starting with the guitar, and from there into a “scratch” vocal – i.e. something to help follow the parts of the song but intended to be replaced with a better vocal take (or composite of the best parts of several takes) later. However, I ultimately liked the initial recording and kept it. I increasingly think that 2-3 takes should almost always be enough for vocals, if the performance decisions (how this or that line should be sung) are already made. This isn’t really counting flubs or false starts, partial takes where all the lyrics completely vanished from your mind, etc. But two solid takes will probably contain a good enough rendition of every line or section, and if it’s not the case you can go in and get another take of just that troublesome spot. And as long as I’m pontificating and pretending to know jack shit: I think vocal comping (compositing) should be done at the largest feasible level of structure – meaning, take the best take of an entire verse or chorus, or at least the best take of a complete line, rather than the syllable-by-syllable painstaking comping that seems to be so prevalent in the modern music industry.
Speaking of austere principles I’ve stolen from better and more experienced engineers, I completely went against one in this song, by indulging in a bit of utter fix-it-in-the-mixery. If that sounds opaque to you: there is a general precept among audio engineers, averred by all and adhered to by almost none, that you should, whenever possible, record things the way you want them to sound, rather than capture something problematic and say, “we’ll fix it in the mix.” I broke this rule when it came to the bass in this song, which I played in an enthusiastic way that felt extremely satisfying but which left me with a recorded sound that had a huge, vaguely flatulent attack to every note, way out of proportion to the sustaining tone. This was particularly true on the choruses and intro – I think I didn’t rerecord it mostly because I liked the sound on the verses, where I’d played softer. I ended up wrestling with the bass tracks quite a bit trying to tamp this down, with compressors with different attack settings (mostly short to try and push down this obnoxious transient), dynamic EQ (TDR Nova) and even a tape emulation (Variety of Sound FerricTDS) on the bass bus.
So that part was a slight misadventure, but my overall experience was super fun. After the guitar and vocals (which were recorded with my Rode NTK tube condenser) my main approach was simply: pick different microphones for different instruments, stick them in front of me, and record the part. I think I used a Line Audio CM3 on the baritone ukulele and my sE 2200a II on the soprano uke – unless that was also the CM3. I need to take notes. I definitely used the 2200a II on tambourine, in figure-8 mode. The last thing I recorded was my self-harmony on the second and third choruses, which I also don’t remember the mic used on. In any case, the recording generally went well, even if I didn’t take notes and screwed up the bass and had to fix it. And besides, Zetalot liked it:
— SK Zetalot (@Zetalot2) January 3, 2018