Studio SOS: Recording Guitar & Vocals Together

The SOS team answer a reader's cry for help, and assist him to improve his recordings of guitar and vocals.

Following the lead of the numerous home and garden make-over programmes that seem to be taking over the television schedules, when SOS reader Steve Graham sent us an email asking for advice on some recording problems he'd been having, SOS Technical Editor Hugh Robjohns and I decided to pay him a visit at home, to see if we could come up with a solution on the spot. Armed with a couple of preamps, an assortment of mics and a bag full of cables, we set off for Bristol where Steve had his studio set up in a small bedroom.

Following a previous enquiry to SOS, Steve had already equipped himself with a pair of CAD M179 multi-pattern microphones, which he was using via the mic preamps of his Mackie 1202VLZpro mixer, into his Roland VS1680. He'd already decided that the Roland mic preamps weren't sensitive enough and he felt the quality of the Mackie preamps was better, but he still wasn't getting the result he wanted. Unfortunately, one of his newly acquired mics had mysteriously stopped working, but, as luck would have it, Hugh had brought along one of the same type (but in a fetching shade of blue), so we were able to continue.

The Problem: Recording Guitar & Vocals Together

On the face of it, the problem seemed simple. For this particular project, Steve wanted to record just his acoustic guitar and vocals together as a live performance, but this was complicated by two factors. Firstly, he tends to play with the guitar fairly high up his body, so there's little physical separation between the vocals and the guitar. Not only does this make balancing difficult, it also opens up the possibility of phase problems, because the vocal mic picks up a significant amount of guitar.

In this kind of situation, the ideal is to space the mics apart by at least five times the distance between mic and source, but here the spacing and mic distance ended up being about the same. The second complication was Steve's playing style, which features some very nice finger picking punctuated by rhythmic damping as he slaps the heel of his hand onto the strings near the bridge. While this sounded fine live, when you came to record it the guitar sound was dominated by loud, deep thumps that peaked 12-15dB higher than the average guitar level.

Steve had tried recording in his usual standing position and, after reading up on recording techniques in SOS, he'd constructed an absorbent corner in the room using two up-ended mattresses. This certainly helped with room reflections, but the close proximity of the mics and the heavy string slapping were still giving problems. He'd started off using the mics in wide cardioid mode and then done a few experiments with figure-of-eight mode, but the improvement wasn't nearly enough. It was at this stage that we decided to lend a hand.

Experimenting With Mic Positioning

After we'd arrived, introduced ourselves and downed the obligatory coffee (Steve also came up trumps with chocolate biscuits, just in case anyone else is thinking of asking us round!), Steve played a couple of songs through to us as though performing live so we could see what we were dealing with. We immediately identified the high position of the guitar as a problem, so we asked him to try playing and singing on a tall stool where the guitar could be positioned a little lower without unduly compromising his playing style. He also tended to sway and move quite a bit when playing standing, which meant close-miking was a problem, so sitting on the stool also helped here, by reducing the amount of physical movement, allowing us to place the mics a little closer for better separation without fear of him swaying out of range.

We patched his mics, still in cardioid mode at this time, into the two SPL Channel One preamps I had brought with me, set with no compression or EQ at first so as to approximate what he'd get if he used the Mackie preamps. The reason we didn't use his Mackie preamps straight off is that I wanted to have the SPL's EQ and compression at my disposal if we needed it.

Here you can see where cardioid microphones were found to give the best results — note the up-ended mattresses in the background, being used for some rough-and-ready acoustic treatment.The guitar mic was set up around eight inches away from the instrument, aimed roughly at where the neck meets the body, while the vocal mic was slightly closer and equipped with a pop shield (laddered, but still effective). The lowering of the guitar, combined with moving the vocal mic to just above mouth level, increased the separation to the point where it was possible to get a good guitar/vocal balance, and the general quality of the recording was good, with no obvious room coloration.

However, that slapping action was still clearly causing problems. At this early stage, we were making sample recordings both to Steve's Tascam DA20 DAT machine and to the VS1680 (to eliminate the VS1680 from our enquiries!) and, though the DAT seemed marginally clearer, the difference was not significant enough to worry us. What did confuse at first though was that the headphone output of the Roland VS1680 sounded noticeably less clear than the headphone output from either the Mackie mixer or the DAT machine, though this turned out to be a problem only with the headphone amp and not with the general quality of the recording. Steve was using 80Ω Beyerdynamic DT250 headphones, and it is possible that the fairly low impedance may not have suited the Roland too well. To solve the problem, we took the Roland VS1680's main outputs through the Mackie's tape return and used its headphone amp to check our playbacks.

Dynamics & EQ Settings

Test recordings made to DAT while setting up revealed problems which needed flexible equalisation and dynamics to sort out, so a couple of Paul White's external preamps that had been brought along were pressed into service to help out.

While Hugh checked the mic positions, I set up the dynamics and EQ sections on my SPL preamps to do something about that thump, so I ended up with around 8dB of cut at 120Hz on the guitar mic plus a couple of decibels of the SPL's Air Band EQ boost (a wide-band 14kHz peaking filter). Then I adjusted the compressor with Steve playing, so that only a nominal amount of gain reduction was showing when he played normally, but the compressor really slammed in when he started slapping. The SPL compressor is an automatic soft-knee design, so I only had to adjust the amount of compression and the make-up gain. From the gain reduction meters, I'd say there was between 10 and 12dB of gain reduction applied during those slaps, but the recovery was so fast that the rest of the guitar playing was unaffected.

The vocal was treated to the barest hint of compression and a decibel or so of Air Band EQ, and the 75Hz low-cut filters were switched in for both preamps (the high-pass filters on the mics were switched out to allow experimentation from the preamps, although we would have had the same result had we used the mic's filters.

Straightaway, the result was far more to Steve's liking, with the thump tamed and the Air EQ adding clarity without harshness. The final step was to pan the vocal track just left of centre and the guitar track to around three o'clock, which produced a reasonably well balanced sound with a hint of stereo spread. No effects were added at this stage. In isolation, the guitar track sounded a little bass light, but when the spill from the vocal mic was added in, the tonal balance was actually rather good.

Decreasing Spill Using Figure-Of-Eight Patterns

An improved setup using the two microphones in figure-of-eight mode.Though we were getting very usable results by this time, both Hugh and I were keen to try using the mics in figure-of-eight mode, because this way it's possible to exploit the off-axis null of the figure-of-eight pattern to improve isolation. To do this, we set up the vocal mic so that it's 'deaf' axis was pointing directly at the guitar body and the guitar mic was set up so that its deaf axis was directly in line with Steve's mouth. Of course a figure-of-eight mic picks up equally from both sides, so to avoid problems from room reflections, we hung a duvet from a mic stand, directly behind the mics. This meant Steve was almost totally surrounded by acoustic treatment of one kind or another, hence the photo looking down on him from above!

Checking out this setup showed we had reduced the amount of voice spilling onto the guitar mic by at least 6dB, and lowered the guitar spill to the vocal mic by about 3-4dB. While this may not sound like a lot, it reduced the levels of phasing when the two mics were combined in mono, and it also made getting a good balance much easier. The reason the guitar spill was reduced a smaller amount than the vocal spill was due in part to reflections from the ceiling, but also to the fact that a guitar is not a point source of sound. The angle of rejection of a figure-of-eight mic is actually very narrow — rather narrower than the radiating surface of an acoustic guitar body — so unless you're recording sound sources that only radiate along a narrow line, in near anechoic conditions, you won't get anywhere close to the theoretical rejection. It is worth paying close attention to the alignment of the vocal mic to optimise the relative angle of the figure-of-eight null and the axis of the guitar body as a few degrees difference can increase or reduce the separation by several decibels. The improvement we achieved here changed the balance, so a little adjustment to the pan controls was required to get the vocals back into the centre of the mix.

Hugh Robjohns auditions recordings through Steve's Mackie 1202 VLZpro mixer — although Steve's VS1680 recorder has a headphone output, better results were achieved by monitoring the main outputs through the external mixer.

As an experiment, we repeated the recording substituting a Sennheiser MKH30 small-diaphragm mic on the vocals to see if it would give even better separation. As it turned out, it did improve the separation slightly, but Steve felt it didn't suit his voice as well as the larger diaphragm CAD M179.

At this stage we had a recording that stood comparison with what you'd expect to hear on a record — it only needed a little additional ambience to complete the picture. Before setting up a reverb however, we decided to try the same recording technique using Steve's own equipment instead of the SPL preamps. Although the Mackie mixer gave us a beautifully clean recording, its EQ wasn't flexible enough to sort out the slapping problem (and neither was that of the VS1680), so we advised Steve to put a good-quality hardware equaliser on his shopping list. Steve already owns a Drawmer DL241 compressor, which should allow him to achieve very similar results on the guitar to those we obtained with the SPLs, without having to buy anything more than an equaliser. As it happened, he'd already half decided to buy a two-channel TLA equaliser because of its hybrid tube circuitry and because it included good mic preamps, so we saw no need to advise him otherwise.

Final Touches

Turning now to the 'polishing' part of the job, Steve had a Quadraverb and the effects processing in the VS1680 available, although we soon discovered that Steve was a little unsure of what some of the reverb parameters actually did. We found the Quadraverb was a little too coarse for this Job, but the Roland effects seemed adequate. After explaining the role of the adjustable parameters, I had a go at adapting a Small Room program in the VS1680 to give a more intimate vocal sound that had space without sounding washy. Steve didn't want a 'steamy' sounding modern reverb anyway, as the sound he was after was closer to that of folk records made in the '70s, when most reverb would have come from a plate or from a natural acoustic.

Once the sound had been recorded as well as possible, Paul White helped Steve to create suitable reverb and compression treatments for the mix within the Roland VS1680 multitracker.

Because there seemed to be no dedicated ambience programs, I thought the best approach would be to crank up the early reflections level to near maximum and reduce the decay density and diffusion to emphasise the early reflections. What we finally arrived at was based on a Vocal Room program with a 1.8s decay time, 80ms pre-delay, medium density (44) and medium diffusion (46) with the early reflection level up at 96. A high cut at 10kHz coupled with 3dB of HF damping (14kHz) rounded out the high end, and by adjusting the reverb level carefully, we were able to get a good sense of space without the reverb being too obvious. The Room Size parameter was set to 30m.

Steve was pretty happy with the sound we had achieved together, but happened to comment that his mixes never sounded as loud as those on commercial CDs, so a whole discussion of mastering and, in particular, the use of dynamics followed. Though the VS1680 includes some mastering programs, Steve wasn't confident enough to adjust the parameters, so he'd experimented with some likely looking presets, but found the results a little too coloured sounding. Also, as these mastering plug-ins can't be used at the same time as reverb, and as we couldn't take the time to do the mixes, we decided to try a regular full-band compressor to see if this could be used to fill out the sound and push up the level.

This was used as a master insert effect and set up with 1.5:1 ratio (the ratio lowest available) and around -20dB threshold. Attack time value was 24 with release time set to 50. The recording was made leaving around 10 to 12dB of headroom, so I guess the effective threshold setting relative to the peak level was somewhere between -5dB and -10dB. As it turned out, this was more successful than expected, pushing up the level and helping to knit the guitar and voice sounds together.

As Steve expressed an interest in having his material professionally mastered, we didn't go any further down this road, but I did offer to master a few tracks for him at my own studio and to explain the processes that could be used. He was particularly interested in the possibility of using valve or valve simulation technology to produce more of a retro sound, closer that of his favourite '70s recordings.