Optimising vocal recordings and tackling some troublesome mixes are on the agenda this month, as the SOS team travel to Paul Shepherd's studio in Nottingham.
Paul Shepherd is one of those musicians who has enjoyed a busy and interesting part-time musical career, but has only just returned to recording after a long break. His previous experience was back in the days of C-Lab's Creator (the predecessor to Apple's Logic DAW software), but today he has Pro Tools 8 LE running on a Mac Pro, which is teamed with a Mackie Control surface, a Digi 003 interface and a pair of Mission 762 passive hi-fi speakers powered from a Denon amplifier. His studio is set up in a small basement beneath his Nottingham home, and measures approximately two metres by a little over three metres. The ceiling height is low enough to require taller people to duck. He records mainly original material along with his daughter, Madeleine, who, despite being only 13, has an enviably strong and well-pitched voice. She also has some specific ideas on how contemporary music should sound.
Paul contacted us because his mixes didn't pack the punch he felt they should, and they also sounded different when played outside the studio — a common problem and one that can have multiple causes.
Over coffee and some very welcome bacon sandwiches, we played some commercial material from Hugh's test CD. Given the size and layout of the room, the results were fairly predictable. Paul had set up his studio facing across the width of the room rather than along its length (lengthways is generally safer in smaller rooms), but given that he'd also managed to squeeze a Roland V-Drum kit and a selection of keyboards into the space, it would really not have been practical to set it up in any other way.
The main problem was that Paul's mixing position was almost exactly halfway between the wall behind him and the wall behind the monitors, and in small rooms with solid walls, experience has shown that this produces a dead zone where much of the bass end is lost. We demonstrated this to Paul by asking him to listen to a track from his normal position and then to move closer to his desk to take him out of the central area. He immediately heard that the bass end was better balanced when he moved forward from the dead zone.
This physical acoustic problem can really only be solved with very extensive bass trapping, which would reduce the usable space in the room dramatically and unworkably. The best we could realistically achieve was to make sure Paul understood the issue and leaned forward when trying to balance the bass end of his mixes. In this kind of circumstance, it becomes really important to check mix balances in other rooms and on other systems. Although this slows the process down, it does make it possible to arrive at mixes that translate well to other systems.
Paul had already put up a small amount of fairly thin 'camera case' foam at various points around the room, which, along with all the equipment and shelving, neatly eliminated any flutter-echo issues. The Mission monitors were mounted quite high on the wall, but the unusual design placed the tweeter in the middle of the front baffle (with the bass driver at the top and reflex port at the bottom), and this arrangement conveniently put the tweeter at around head height. By chance, this produced surprisingly good results, given the room's size, low ceiling and thick, solid walls, though both Hugh and I suggested that Paul check out the Focusrite VRM Box, which emulates various speaker types over headphones. While I wouldn't advocate using a VRM Box for making all mix decisions, it works exceptionally well as a secondary reference for checking that your mixes are likely to translate well to other systems.
For recording vocals, Paul uses the corner nearest to the door, with the door itself covered in more of the thin, camera-case‑style foam. His Rode NT1A mic is teamed with a Studiospares reflection screen, plus a mesh pop filter, and he was getting reasonably dry results. We felt we could improve the situation, however, by adding some acoustic foam to the side wall, so that both sides of the corner behind the singer would be less reflective at vocal frequencies. The low ceiling was also a worry, so we decided to fix a two-foot-square tile there too, and as we had a few two‑inch‑thick Vicoustic pieces left over from a previous Studio SOS, we decided to use those.
Alhough they lowered the ceiling headroom by a further two inches, the foam tiles reduced ceiling reflections, and provided a more comfortable experience for any taller people entering the room! Paul was happy for the tiles to be glued directly to the wall, so out came the spray glue and, 10 minutes later, the job was done. Hugh also observed that Paul had positioned the mic rather far back inside the curved mic screen, and in our experience the sound can become coloured unless the mic is brought forward to be roughly level with the front edges of the screen. Again, a couple of quick adjustments by Hugh remedied this, and the result was a noticeably more neutral-sounding vocal area. However, the screen and mic combination was mounted on a mic stand with a relatively small round base and, as a result, the whole thing was decidedly unstable, so Hugh recommended purchasing a stand with a tripod base, which would be a lot more steady.
Having addressed the more mechanical problems, it was time to look at one of Paul's mixes to see what other issues he was experiencing. He'd identified the bass sound and vocal treatments as particular areas of concern, but had only a limited number of plug-ins at his disposal: the ones that come with Pro Tools LE, plus a JoeMeek compressor and Izotope's Ozone Lite, which allows the user to access a range of presets, but not to edit them. Paul also confided that he'd tried to use level automation to keep the vocal level even, but hadn't used any compressor plug-ins. The only vocal treatment was from the Air Reverb plug-in that comes with Pro Tools.
Playing the track revealed a very capable arrangement with a good choice of sounds and a well-performed vocal, though the vocal sat rather too far back in the track and the relatively unsophisticated Air Reverb didn't really do it any favours. There was also a lack of focus to the low end. Paul admitted to mixing with his monitoring cranked up to very high sound levels — possibly as a natural response to the inherent lack of low end due to the room's acoustics — and this would certainly account for the relatively submerged vocals. Paul had used Pro Tools LE's Strike drum instrument for all the drum parts, and a sampled bass guitar, again from one of the included plug-ins. He'd added a little electric guitar of his own, via a Line 6 Pod XT, and was using one of the Izotope Ozone presets to process the final mix. He felt that the overall sound was slightly abrasive and lacking in punch, but didn't know how to remedy it.
As has been explained in many of our Mix Rescue articles, not all bass instruments need to produce large amounts of deep bass, as the human ear finds it quite difficult to separate sounds at the lower end of the frequency spectrum. In this mix, the kick drum was providing plenty of weight, but it was battling for space with the sampled bass guitar. Our solution was to roll off some low end from the bass guitar, and also to boost it at around 230Hz to emphasise that area of the bass sound that would be most audible on a limited-range system, such as a portable radio. We also rolled off some of the extremely low frequencies from the kick drum, as they were simply eating up headroom for no good reason. In most cases, there's doesn't need to be much going on below 40Hz, even on kick drums, so the trick is to use a low-cut filter, and then move the frequency up until you can just hear the sound being affected, then back it off again slightly.
Strike allows various aspects of the drum‑kit sound to be balanced, so to clean up the drum mix we reduced the room mic's contribution significantly and pulled the overhead mic level down slightly. Strike also has a fader that emulates the infamous SSL talkback mic compressor used on so many drum tracks in the '80s. This is a very 'pumpy' compressor that, when added to the dry drum sound, gives it a lot more attack and attitude. This is essentially parallel compression built into the Strike plug-in, and it can be used as subtly or unsubtly as you like. We found that a fairly subtle application gave the drums the right amount of edge.
For the bass part, we attenuated the lows below 120Hz, using a shelving filter. It's also a good idea to high-pass-filter vocal and electric guitar tracks, as these can include low-frequency elements that are not part of the wanted sound and just muddy things up. For example, a vocal mic can pick up low-frequency air blasts, even with a pop screen in place, while the electric guitar generates low frequencies when the strings are touched while palm-damping, and also sometimes during the picking phase of the lower strings. The lower-mid range of a pop mix can get very congested, so any unnecessary lows that can be cut will help keep this vulnerable area clear. After tidying up the low and mid-range frequencies in this way, we turned our attention to the vocal sound.
Aside from being mixed a little too low, the vocal part wasn't as even-sounding as it could be. Given plenty of time, I'd generally use automation to level out as many level fluctuations as possible, then feed the vocal to a bus with a compressor inserted, so that the compressor comes after the automation. In my own studio, I often use Wave's Vocal Rider instead of automation, as this does a great job and attends to the fine detail of level changes within individual words very efficiently, although I find that it sometimes helps to precede this with a low‑cut filter, to prevent 'near‑popping experiences' from triggering gain reduction.
In this case, though, we bypassed the automation that Paul had already created, so that we could demonstrate what was possible using a compressor. As is the case for many readers, he was a little unsure as to how compression should be applied, so on the occasions he had used, it he'd dropped in a preset without realising that changes to the threshold or input level still need to be made to achieve the desired amount of gain reduction. For our vocal treatment, we used the JoeMeek plug-in to give us four or five decibels of 'squash', and then followed this with the included Maxim plug-in, which is essentially a type of limiter, to take care of any excessively loud peaks.
That gave us the necessary 'up-front' and dense sound, so then it was down to looking at reverb and delay treatments. The original choice of Air Reverb produced a rather dated sound, with the reverb being a little too obvious, so we switched to D-Verb and used the Plate setting with around 30ms of pre-delay. The reverb time was reduced to around one second, and a second aux send was set up to feed an instance of the Stereo Delay plug-in. This was sync'ed to the song tempo, with one repeat at around 300ms and the other at around 600ms. Feedback was set to about 30 percent. By combining the reverb and delay at a level low enough that it was not too obvious, but was still 'missed' when bypassed, we made the vocal sound much more focused and contemporary. When solo'ing the vocal, the delays became very obvious, but in the context of the mix they were almost subliminal. This mix of delay and more reserved reverb treatments is very popular at the moment and helps keep the vocal at the front of the mix.
In fact, Paul had already employed some more obvious delay treatments for selected phrases in his mixes, but rather than automate the send level to a delay plug-in, he had adopted the technique of copying the final syllables of some vocal lines to a new track, and sending that to his delay plug-in, to add delay precisely where he wanted it. This approach is just as valid as automating the sends, of course.
After we'd balanced the vocal a little higher in the mix, our final tweak was to use some high-shelving EQ to add a couple of dB of boost above 6kHz, to lend a sense of 'air' to the sound. We demonstrated how, by applying maximum boost while adjusting, it was possible to move the frequency control and hear at what point the EQ changed from adding high airiness to upper‑mid harshness. Once you find the best corner frequency, you can adjust the boost level.
For mix processing, we found some smoother-sounding Ozone presets than the one Paul had chosen, but we felt it would be more useful to use separate plug-ins, which we could adjust, so that Paul could follow the process and also save the result for later experimentation. For EQ, we chose Pro Tools LE's included four-band equaliser, and created a very subtle 'smile' curve using 2dB of boost at 88Hz, 1.5dB of cut at 240Hz and a generous 4dB of shelving boost at 8.5kHz. This was followed by low‑ratio, soft-knee compression of 1.4:1, with the threshold adjusted to give around 4dB of gain reduction. Then we used the Maxim limiter to shave another 2dB off the peaks. This combination produced a punchy sound that didn't suffer from the harshness of Paul's earlier mix.
The second song required a different vocal treatment that would add life without sounding obviously 'reverby'. For this, we chose D-Verb's Medium Ambience program, opened up the high-cut filter to keep it bright, and then added around 30ms of pre-delay. In conjunction with similar compression to that used in the earlier song, this added a subtle density to the sound and eliminated the dryness. We also took the opportunity to show Paul how a higher reverb level and longer pre-delay time could create a vintage, slap-back doubling effect.
For the drums on this song, Paul wanted a gated-ambience drum sound but didn't know how to go about achieving it. We used PT LE's Non-Lin reverb for this, and though it doesn't deliver the best gated drum sound in the world, it does a perfectly good job. Essentially, it produces a dense burst of early reflections that remain at a fairly constant level, before cutting off abruptly. Getting the balance right is the key to the success of this sound, and adding in some of that talkback-mic compressor really helped achieve the desired aggression.
That left only the electric guitar sound. Paul's daughter, Madeleine, likes bands that use a fairly dirty rhythm guitar sound, so I suggested we listen to some of her commercial choices, as well as some Green Day songs, to see how the guitar should sound. I usually opt for miking a guitar amp, but in Paul's small room a DI approach seemed more practical.
I set up a couple of Pod sounds that we could combine to get the desired result. Paul wanted what he described as a 'chuggy' sound, but I know from experience that overdriven parts can get very messy unless you take care with both the sounds and the arrangement. I set up a Soldano amp model to give a dense overdrive with plenty of low end and a bit of mid-range scoop, for almost a classic metal tone, but with a little less presence. I then found a Vox AC30TB model that had enough drive to produce ringing chords with a nice dirty edge, but without losing definition.
The idea was then to double the part, using the AC30 sounds for the full chords and the more dense Soldano sound for playing just the bottom couple of strings (in this case the root and fifth) of the chord. I tried this using Paul's rather nice Rickenbacker and it sounded pretty good. However, you do have to play 'chuggy' parts appropriately, with plenty of attack, followed by releasing the left-hand finger pressure slightly between strokes to afford a little damping. Without the damping, you tend to get a more continuous sound that can be messy, whereas appropriate damping helps reinforce the chugging rhythms.
We discussed a few other guitar strategies, such as doubling the part while playing only the higher strings of the guitar on the first pass and the lower strings on the second, and also layering two parts playing different inversions of the same chord. If in doubt, recording an unprocessed dry sound alongside the processed sound is always safest, as it allows you more flexibility when mixing. Often layering a dirty and a cleaner version of the same part will save the day if things are getting too messy, as mixing in the cleaner sound has the effect of adding some definition back into the dirty version. These and other guitar-recording tricks will be explored in greater depth in a future article dedicated to the subject.
Our final piece of advice to Paul was that he consider upgrading to Pro Tools 9 or 10, as these versions offer several improvements, including plug-in delay compensation. This could be an important consideration, as Paul was feeling tempted by a UAD2 DSP card, and the effects that run on these incur greater latency than non-DSP plug-ins.
We left Paul with plenty to think about, but he seemed to have found the exercise useful, and we both look forward to hearing his future mixes.