Page 2: Crafting Perfect Pop Backing Vocals

Editing

Getting multiple tracks of harmonies to really work together can be a lot of fun, but it's usually a lot of hard work. Repairing a track or two with micro-editing is fine, but while the techniques outlined below can be applied to lazy performances, having to micro-edit a multitracked block chorus will quickly have a debilitating effect on your will to live! The results aren't likely to be as good either, so having decent recordings to begin with is essential! On a more positive note, with quality material and a little effort, great results are achievable using the basic tools available in any DAW software. The essential rule is to establish a reference against which edits will be made, and with double-tracks and simple harmony lines the reference is, obviously enough, the lead vocal.

Headphones are a great help in spotting timing problems — just pan two parts to opposite ears and such problems will stick out like a sore thumb! When you hear problems, zoom in on the waveforms in your DAW and edit parts to bring them into line. Although it isn't always necessary, note lengths can also be manipulated, often by fading parts so they end early, rather than using time-stretch tools unnecessarily.Start by placing the reference lead and the vocal to be edited on adjacent DAW tracks, just to provide a visual guide, then solo both tracks, and pan each hard left and right at a roughly equal volume level. It's not a bad idea to use headphones for this particular job.Such a forensic approach can be brutally revealing, to the point where it's often not much fun, but in the long run it's the best route to perfectly tight tracks and a professional result. Any timing issues will be quickly revealed by the extreme panning: good sections will flow past easily, while problem parts will jar uncomfortably. Once you encounter such a clash, zoom in on the waveform in your DAW. Most timing edits can be accomplished simply by separating out and selecting wayward words or syllables and nudging them into alignment with the reference. Make sure you add appropriate fades to the snipped events and the audio either side, just to avoid clicks — though occasionally you may need to stretch or shrink sustained notes to be really precise. As a rule, if the BV is noticeably early, deal with it as a matter of course. Conversely, a limited degree of lateness is often acceptable, unless hard-sounding consonants cause an unpleasant stuttering effect.

Headphones are great for sorting out the timing of your massed vocal tracks. Use one part as a reference for timing — for example, the lead vocal if you're doubling that — and hard pan that to the left ear. Then hard pan the track you're editing to right. This makes any annoying timing discrepancies much more obvious, so the job of making things sound in time becomes much easier.Once everything's sitting tight, use level automation to deal with loud or quiet words and syllables. Strong breath intakes are often best dealt with by drastic level reduction rather than complete elimination, which can sound artificial. If you plan to do the edits and then bounce them, you can use the track's channel fader, but if you tend to mix and edit in the same project, you want to apply the level changes before any further processing — for example, managing the levels of clip envelopes, or automating a gainer plug-in at the top of the processing chain. Another option in this case is to use a dedicated breath-attenuation plug-in such as Waves DeBreath — but do make sure that you listen to the results as carefully as if you were editing or automating manually. Clicks and pops should, of course, be cleared completely, along, where possible, with any other extraneous noises such as headphone spill.

Breath sounds contribute to the rhythmic feel of parts, and will usually require editing to bring them into time, and some degree of attenuation. For the latter task, you could choose to use automation or a dedicated plug-in such as Waves DeBreath.I find that automation is also the best way to deal with excessive sibilance, though approaching the job in this way may be time-consuming. Most standard DAW de-essers seem sadly lacking, often colouring the sound while missing some esses and making lisps of others. So unless you have access to a high-quality proprietary de-esser such as Eiosis E2, be prepared for a little manual labour: it's well worth it. The good news is that you don't have to de-ess every track of a multitracked vocal, as the process can usually be applied after the parts have been bounced down to a single track.

At this point, you may as well deal with tuning discrepancies too — but before reaching for a plug-in, check that any problematic notes are not performed perfectly earlier or later in the track, in which case a quick cut-and-paste may well do the job. If your track is generally OK but not-quite-on-the-money, the pitch correction plug-in bundled in your DAW will probably deliver an acceptable result, albeit at the expense of a little sonic fidelity. More drastic problems will need the attention of dedicated software such as Celemony's Melodyne or Antares' Autotune, either of which is a must-have purchase for anyone unfortunate enough to have to deal with poor tuning on a regular basis. As a bonus, in addition to being excellent corrective tools, both are useful as creative devices for generating new harmonies and effects.

Multitracked vocals that do not follow the lead vocal in pitch can still be edited using the 'master take' of the part as a timing reference. Again, using the solo/left/right method will lead to the tightest results, and breath intakes are normally best reduced in level rather than totally wiped clean — although any that are blatantly out of time with the master should be deleted.

If multiple parts are forced to be perfectly in tune using software such as Melodyne or Cubase's VariAudio (pictured), you'll probably create unnatural-sounding 'phasing' effects, and also lose all the subtle sonic richness you wanted from the multiple parts in the first place! But remember that you can also use such tools to create disposable guide parts to help your singer if they're having difficulty with the harmony parts you've written!Tuning multitracked vocals needs extra care, as simply strapping multiple instances of a pitch-correction plug-in across a bunch of tracks sung by the same vocalist can lead to nasty phasing problems. To avoid this, adjust the absolute pitch reference of successive plug-in instances by a few cents flat or sharp. Remember, much of the glory of multiple vocals is due to the subtle but inevitable tuning discrepancies between takes, so completely ironing them out defeats the whole point of the exercise!

Processing

Pop backing vocals will nearly always — no, make that always — benefit from EQ and compression, which can be applied in two doses either side of bouncing down. For the pre-bounce processing, try to err on the safe side, using it as a corrective rather than creative measure. Your aim is to create a solid, problem-free bounced vocal suitable for more adventurous treatment later. Here's how to go about it.

These two EQs were used on similar parts in the same track. While the settings are obviously different, tuned as they are to individual parts and the rest of the track, you'll see that in both cases quite radical high-pass or low-shelf filtering has been used. Such filtering, at both the high and low end, is not at all uncommon for backing vocals.First, remove any low-frequency noises and rumbles with a low shelving cut at about 150Hz, then use further EQ dips to gently deal with any tonal shortcomings in the vocal, such as muddiness or honkiness. Generally, try to use quite narrow Q settings for cuts like this, though you can start wider to find the problem frequency, and then narrow it as you home in on the offending frequencies. A good dose of compression can then be prescribed to bring some 'cohesiveness' to the sound and smooth out any major level jumps, but without squashing the dynamics completely. Use a gentle-ish ratio: as a broad rule of thumb, the gentler the ratio, the lower the threshold can be, but you must use your ears when setting the threshold, as this parameter is always programme-dependent.

Once you've arrived at a useful combination of settings, make a template of the channel strip so you can apply it to the other similar tracks. As with any presets, some tweaking of this template will almost certainly be needed for most parts, but at least you'll be working from a good starting point, and this will save you time.

Bouncing Down

Despite its ancient origins, bouncing down is still a valuable process, whether it be as an aid to mixing or project navigation, or simply a way of freeing up CPU or RAM resources. Thirty-two or 48 tracks are a lot easier to work with than 173! Also, given the susceptibility of automation data to the occasional gremlin, even a single track can be worth bouncing, especially if it also contains multiple edits.

In the case of our multi-tracked three-part harmony, which may easily involve 24 or more tracks, the aim is to emerge with three stereo tracks, one for each harmony, so that each part can still be manipulated at mixdown. Before committing, check each element's stereo placement and that the overall image is as wide as possible, as this is the major area in which the finality of the bouncing process can lead to compromise later on. If the mix isn't wide enough, it can only be increased by stereo-image manipulation, which will sound artificial. A fully-wide, hard left/right image can be narrowed without sonic compromise, though, so that's the best route to take. Ensure that no reverbs or delays are included in the mix (these are invariably best added later, in the context of a mix) and that any master bus plug-ins are bypassed.

Mixing

The labour involved in editing and bouncing down bears fruit at the mixing stage: you should now have a minimal number of great-sounding tracks, yet still enough flexibility for creative processing. Bounced vocal tracks also make great raw material for samples, if chopped into snippets, reversed or otherwise effected and used at different points in the song's arrangement.

If you want to hear an example of successful bracket EQ'ing and heavy compression of backing vocals, you could do worse than listen to Adele's 2011 smash hit 'Rolling In The Deep'.However, the main function of bounced tracks is to facilitate mixing, with the intention of getting our backing vocals to do their work without crowding out the lead vocal or obscuring instrumentation. The single most effective technique is to 'bracket' the BVs into their own space with some strong low-cut shelving, plus taking a little off the top, before applying some severe compression. As a rule, backing vocals can handle shockingly hefty portions of squash, as there's rarely any need for one particular word to be dramatically louder than another. This EQ/compression combination will quickly separate the BVs from the lead vocal part, and the extent to which it is done — which can be huge — is purely a matter of taste. If you want to hear this in action, check out the chorus of Adele's 'Rolling In The Deep' (the biggest hit of 2011).

Further separation from the track and/or lead vocal can be achieved by adding a second compressor, with its side-chain keyed by another element of the mix which may be in danger of being obscured. As an obvious example, you could key the second BV compressor with the lead vocal signal, to achieve 3dB of gain reduction in the backing vocals whenever the lead itself is present.