Some creative thinking and a little DIY pay real sonic dividends when recording four vocalists.
Last month I discussed my recording, on location in the theatre of a local music school, of Spektakulatius' latest album. An unusual aspect of the band's line-up is that it features four singers, who share out the lead and backing duties between them, sometimes switching roles in the same song. As we were on a tight schedule (three days to record 14 songs), I was keen to capture the full band and all the vocalists simultaneously, but I also wanted the ability to edit each vocal part independently, and even to comp their parts between different takes if necessary. Fortunately, we had the luxury of a separate room where we could record the singers in isolation, so spill between the vocals and the band wasn't a concern. But that still left the challenge of how best to reduce spill between the singers themselves.
On the one previous occasion I'd worked with this band, we'd been using the main function room of a converted pub as our recording space. It had a little raised stage along one side, which the new owner had partitioned off with fibreboard and plexiglass for use as a video control room, and we ended up using that space as a vocal booth. Though workable enough, there were several drawbacks. Firstly, the space was pretty small and had no acoustic treatment, which gave it an unflattering 'medium closet' ambience. I did my best to damp this by hanging blankets and duvets, but we didn't have unlimited supplies of those so this only reduced the boxy resonances of the enclosed space a certain amount and left me with more EQ work to do at the mix than I'd have liked.
The second problem was that the booth's long, narrow shape forced me to set the singers up side by side; I was only able to separate them with fairly flimsy DIY cardboard-and-blanket partitions, which weren't a huge amount of help, to be honest. Also, the vocal mics I was using were cardioids, which only attenuate by about 6dB at 90 degrees off axis anyway, and the bleed between the singers proved problematic when mixing a few numbers. Again, I found ways around this at mixdown, but I didn't fancy repeating that extra work this time if I could avoid it. Finally, as the singers were standing in a line, with partitions between them, they had poor visual contact, making it trickier for them to maintain tight ensemble timing and match the phrasing of their lead and backing lines.
To reduce these problems for this recent session, I decided to try creating a bespoke baffle, using wood and Perspex to get better spill rejection (especially at high frequencies, which I knew I might want to boost at mixdown), while maintaining better sight lines between the singers. A major design criterion was that I didn't want to fully enclose any singer or have them between parallel hard surfaces — that would have been a recipe for nasty tonal colorations from boxy booth resonances or metallic-sounding flutter echoes. I settled on a cross-shaped construction (seen from above), with one singer in each quadrant facing the centre of the cross. Not only did this allow direct sight lines between all the singers without anyone having to turn from their own mic, but it also positioned each singer at least 90 degrees off-axis to all the other singers' mics, improving spill-reduction. Minimising spill and room sound from other directions would be the job of blankets and duvets slung over the top of the baffle and behind each singer.
So far, so good — in theory. In practice, there were additional practical considerations. The first was that I wanted to make the structure around 225cm high — tall enough to provide good separation, but not so big it wouldn't fit under a normal domestic ceiling — and with each 'wall' of the cross 120cm long, but I knew there was no way I'd be able to transport such a large bit of hardware in my car. Thus I had to design the baffle so that it could be flat-packed and then built on-site. Furthermore, I wanted it to be fairly straightforward to build without tools, so that I could enlist helpers to speed up the building process. Another perennial project-studio consideration at the back of my mind was how I could make the baffle reconfigurable for use in future sessions, with different isolation requirements. My response to both issues was to make the baffle design modular...
I constructed each wall of the cross out of four 5mm-thick panels, each 120cm across. Three of those measured 65cm high (two made from MDF, one from transparent Perspex) and the fourth (MDF) measured 30cm high. These four panels were stacked on top of each other, edge to edge, with the smaller MDF panel at the very top, and they were held in place by vertically oriented lengths of aluminium profiling and M6 bolts (with matching wingnuts for hand-tightening). Because the three main panels were all the same dimensions, and the bolt holes were drilled in the same positions in each, it allowed me to place the one Perspex panel at whichever level was most suitable: the top position made most sense here, but had I been recording seated performers the middle position would have been better. I would also be able to use more than one Perspex panel in a wall, or remove panels to make half-height baffles.
The other element of modularity was that, by using L-shaped aluminium profiles, I could connect the four walls of the baffle in layouts other than the cross configuration Spektakulatius were using: a W‑shape, for instance, or a larger V‑shape, or all in a long line. I could also use fewer walls at once, or combine them in different ways, for example to create two V‑shaped baffles.
With this basic plan in mind, it was time to build something and discover whether it would work in practice. Fortunately, my local DIY shop has a free wood-cutting service and I was also able to order the Perspex pre-cut to size, so implementing my idea mostly involved just hacksawing a load of aluminium profiles to the correct length and drilling bolt holes through the profiles and panels. I quickly discovered that drilling the panel holes wider than the profile holes made a lot of sense, because otherwise the holes wouldn't reliably line up — a damning indictment on my own manufacturing tolerances! I also realised that I could cut each length of profile into two sections for easier transport, as long as I staggered the joins between them on opposite sides of each wall of the baffle.
Once a prototype was standing in my living room, it occurred to me that the walls of the cross were only being held in position by friction against the floor. This meant that if anyone were to lean against one, there was a risk that the panel would bend out of shape — and potentially crack the Perspex in the process. In addition, a duvet suspended across the top of the cross would inevitably droop into each quadrant, even if I stretched the quilt's corners over each wall with bungee cords. So I decided to add braces across the top of each quadrant between the ends of the walls. Some simple wood battens worked well, with hinges screwed to their ends so that I could fix them to the profiling at those points with the same M6 bolts I'd used for the main structure. Finally I had something that looked like it might do the job.
The spill was quiet enough that I could use Melodyne to tighten any tuning I wished, without the pitch-processing mistracking.
But how well did it work in practice? We'd already cleared the first hurdle, because the whole shebang packed down to about the size of single project-studio bass trap, and fitted into the car just fine along with all my other gear. Also, because there were MDF and Perspex panels of the same size, I was able to sandwich the Perspex between the MDF to protect it from scratches during transit.
Building the baffle on-site also presented no problems — especially for me, because I was able to delegate the task entirely to my friend and fellow engineer Herb Felho (www.mixxxster.com), who had agreed to help me out on the session! I'd labelled all the panels and pieces of profiling to indicate how they were supposed to fit together, and I'd printed out a picture of the finished product as it had looked in my living room, so I was able to get Herb up to speed on the construction method in only a couple of minutes, and he could then beaver away putting it together without any further input from me.
We'd found some theatrical scenery supports tucked away in a cupboard, which were ideal for hanging up duvets behind each singer (otherwise I'd have had to use mic stands for this), and finally I sealed all the joins between the MDF and Perspex panels with gaffer tape to ensure an airtight barrier.
I'd done a few audio tests while prototyping that seemed pretty clean-sounding, and so it turned out on the actual session too. I'd been a little concerned that early reflections from the surfaces to either side of each singer might cause problematic comb-filtering side-effects, but this turned out not to be an appreciable concern. Perhaps it was because there were two equally strong reflections combined, or because the microphone's off-axis rejection reduced their level sufficiently, but I noticed no obviously unwelcome tonal character being imposed by the baffle.
And, of course, the most critical question after all that effort was whether the spill levels were sufficiently low — and I'm happy to say that they were! You could still hear the other singers if you soloed each of the mics, but the level and tone of that spill was such that I was able to edit any of the tracks freely without obvious chorusing or doubling side-effects between any singer's close mic and their spill on the other singers' mics. The spill was also quiet enough that I could use Melodyne to tighten any tuning I wished, without the pitch-processing mistracking. But don't take my word for it — check out the audio examples I've posted on the SOS site at https://sosm.ag/session-notes-0819-media and judge for yourself!
So, all in all, my little DIY baffled turned out pretty successfully, and cost only about £350 to build (less than $450, including tax). It's still a work in progress, though, because I've learned several useful lessons from the build process, beyond the fact that I can't drill straight! The first misjudgement I made was in using cheaper flat profiling for the outer edge of each wall of the cross, because I assumed incorrectly that the three-layer aluminium-MDF/Perspex-aluminium structure would be stiff enough to keep the baffle fairly rigid. In reality, though, the baffles still had more flex in them than I'd have liked, so for the next session I'll be using L‑shaped profiling throughout, since this shape is inherently much more rigid.
I'd also given insufficient thought to the singers' sheet music requirements — I've rather got out of the habit of thinking about this, because all the vocalists I've worked with most recently have sung entirely from memory. Because of the comparatively restricted space within each quadrant of the baffle, setting up a music stand was a little challenging, so three of the singers ended up just taping their lyrics to the Perspex. Although this wasn't a bad workaround, it didn't put the sheets in the most ideal position for each singer, and also interfered with the sight lines between them all. For this reason, on my next session I'm going to try using bulldog clips to suspend any sheet music from a string suspended across each quadrant, a bit like laundry on a drying line. Hopefully this'll keep the sheets on axis to the singers and at an adjustable height below the sight lines, while at the same time avoiding introducing any further reflective surfaces around the singer.
In a similar vein, it turned out to be a good thing that the duvets I'd rigged up around the baffle weren't too closely fitted, and that there were plenty of lights in the room, otherwise the singers might well have had too little light to read comfortably. This wasn't something that even crossed my mind during the planning stage, but I'll make sure to take some little battery-powered LED cliplights with me next time just in case.
Project-studio recording is often a bit of an experimental process, because each session has its own unique demands. This isn't a bad thing, though, because every time you push yourself out of your comfort zone you come away with additional knowledge and experience that helps you continuously improve your results. However, as this month's column hopefully demonstrates, this attitude doesn't just apply to gear settings and mic techniques — a bit of simple DIY can also pay serious sonic dividends.
This month's column features the band Spektakulatius, comprising Thomas Göhringer (drums, percussion, vocals), Markus Braun (upright bass, bass guitar), Christian Bolz (guitar, sax, flute), Florian Blau (keyboards, accordion, mandolin), Martina Fritz (vocals), Aysun Idrizi (vocals), Ralf Meiser (vocals), and Christian Steiner (vocals, guitar). Their latest album, About Christmas, is available from their website.
You can download a ZIP file of hi–res WAV and quick-play MP3 audio examples in the righthand Media sidebar or use the link below. Or visit the Audio Examples page at https://sosm.ag/session-notes-0819-media
Download | 33 MB