Musicam antiquam præsento. Received wisdom has it that you can’t “fix it in the mix”; this is an exercise in doing precisely that. The cover of Marian Call’s In the Black that Heather and I did a while back (see my post here) was tracked in late 2011, but I was unhappy with the mix. I have subsequently learned a lot, so I thought that it would be an interesting exercise to see whether I could fix it up without doing any additional tracking.
Step one was to get the original tracking out of Cubase. Strictly-speaking, my exports weren’t stems insofar as most of them were in mono with no effects or automation, but in terms of approach, I built “stems” for each mix element—so, for example, where I had previously created a comp “track” across several tracks, the several were boiled down to one and then exported with appropriate silence from 0:00 to the beginning of the clip. This took quite a while, and the result was around thirty quasi-stems that I could import into Reaper.
I was then able to clean up some of the tracking and start fresh with no (or very few) effects. For example, the guide guitar track was audible in the mandolin stem, and a combination of a high-pass filter and some gating eliminated enough of it to make it useable. By contrast, the digital clipping on what had been the main rhythm guitar was unacceptable, and those tracks had to be ditched completely; fortunately, I had tracked several guitars and two of them were tracked at lower levels that were mercifully distortion-free. And then there’s some stuff in the middle: All of the guitars were recorded with a dynamic mic, and while all the EQ in the world won’t make it sound great, it (inter alia) can make it prettier than it was. Several levels of compression of varying levels of transparency also helped reign things in.
So: Can we fix it in the mix, Batman? The answer is a qualified “yes.” I could probably improve this mix (it seems a little toppy, I think), but the purpose of the exercise has been met: Some things can be fixed. Some can’t. This mix is a significant improvement; it is clearer and more detailed. Much of the mud and mush that blanketed the original mix is gone. But not all of it, and there is audible digital clipping in places that was impossible to eliminate; this is the fault of the tracking engineer, who, being too inexperienced to have yet realized that the noise floor in digital recording is so low as to be virtually irrelevant, tracked too hot a signal. The tracking engineer also tolerated excessive noise bleed and performances that were looser than ideal; some of this can be “fixed” (i.e. hidden) in the mix, but most can’t be.
The bottom lnie is that this exercise vividly demonstrates what I had already learned since then: The first duty of the tracking engineer is to get clean takes—with plenty of headroom—of every musical performance that will be in the final mix.