RECORDING VOCALS ON A COMPUTER

Recording to a computer can make it easier than ever before to produce a great vocal sound — but there are also pitfalls to watch out for...

One of the most common questions we are asked here at SOS is still "how can I achieve that 'produced' vocal sound?" Now that so many of you are recording directly to computers rather than to tape or dedicated digital multitracks, I thought it might be an idea to explore vocal recording in the context of a typical computer‑based studio.

First Steps

You're going to need a quality capacitor mic to produce the best vocal sound, and these tend to require phantom power. An audio interface that provides a phantom supply at the full 48V, such as the rackmount box that comes with the Digi 001 system, is thus ideal for computer‑based studios.

The first issue with a computer system is how to get a mic signal into it. Typical soundcards have only line inputs, or, if they have mic inputs at all they're often designed for use with consumer‑quality microphones. Thus, you're always going to need either a mixing console or a separate mic preamp to interface studio‑quality mics.

If you're using a mixer, it is a good idea to establish as short a signal path as possible. This entails using the mixer's channel direct outs, if fitted, or even taking the signal feed from the channel insert send. You can also feed signal out of the mixer via an unused pre‑fade aux send, which is an effective way of using a simple 'something‑into‑two' stereo mixer for simultaneous monitoring and recording. In general, the more unnecessary circuitry you bypass, the cleaner the signal will be.

As regular readers will no doubt be aware, a capacitor microphone will produce the best vocal results in the majority of situations. To use one of these you'll need a mixer or preamp with phantom‑power capability, or possibly a separate phantom power supply box. A few of the better audio interfaces with built‑in mic preamps have phantom powering, but if you plan to use one of these, make sure the full 48 Volts is supplied, as many capacitor mics either fail to work altogether or deliver a reduced level of performance on lower voltages.

Mic positioning for vocals is relatively simple. The main things to remember are to keep the mic away from walls and corners, and to hang up absorbent material against the walls if the room sounds too 'live'. Always use a pop shield between the mic and the singer, and work around nine inches from the mic. My vocal room doubles as a store room, and it seems that the more rubbish I accumulate in there, the better the acoustics get, presumably because the sound is scattered rather than bouncing back from unobscured flat walls. Avoid small square rooms with hard walls, as these can sound very boxy.

Pre‑processing

Plug‑ins are invaluable, but there's still a lot to be gained from using outboard processors before the vocal signal gets inside the computer, especially voice‑channel units that conveniently combine several problem‑solving features into one box. SPL's Channel One, for example, offers a mic preamp, a decent compressor, pleasing analogue EQ, and the option to add a digital output.

When you work largely or exclusively on a computer, signal processing tends to become something you do with plug‑ins, but it's actually useful to compress vocals with a conventional outboard compressor at the analogue stage, before they enter the soundcard's input and pass through its converters. This makes the best use of the dynamic range of the converters. However, it's best to leave room to add more compression later with your plug‑ins, by erring on the side of under‑compression. To do this easily, set up the amount of compression on the outboard unit that seems right for the vocals, then adjust its compression threshold to produce around 3dB less gain reduction than that 'ideal' setting.

Whether you record at 16‑bit or 24‑bit resolution is up to you, but my experience has been that, for pop music vocals where the signal has been compressed to some extent prior to being digitised, 16‑bit is perfectly satisfactory, even for serious projects.

Normally I advocate leaving EQ until mixing, but there are some analogue equalisers that simply sound nicer than plug‑ins, so you may want to apply some EQ on the way in. I use an SPL Channel One for all my computer recording, because it combines a clean mic preamp with a nice compressor and a good‑sounding equaliser, which means that I can do all my processing in one place. Usually a little high‑end 'air' EQ combined with low‑cut filtering, is all that's needed. (If you do choose to use EQ on the way in, as with compression you should take care not to overdo it.)

I recently fitted a digital output card to the Channel One, which makes connecting to a computer very easy. If you're using a budget soundcard with less‑than‑great converters, a preamp with a digital output is a good way to get high‑quality audio into the system, provided that the soundcard has the appropriate digital input.

Setting Up To Record

Once you've tried 'multiple take' audio loop recording with your sequencer, you may find yourself wondering how you managed without it. Here you can see Cubase's (left) and Pro Tools LE's ways of doing the same thing. After recording several takes in Cubase's Cycled Record mode, the takes appear on different lanes in the Audio Editor, but play back on the same audio channel. Mute takes to audition. Here, sections of several takes are in the process of being cut to compile a finished take.

Depending on your choice of audio interface or soundcard, you may have enough separate outputs to enable you to set up a foldback mix using the pre‑fade sends on your sequencer's virtual mixer. If you only have a stereo output, the singer will have to make do with hearing what you hear in the control room. In practice, I tend to send the singer the control‑room mix anyway, as this can be rebalanced to their taste without affecting my ability to record them. Having a headphone amplifier within reach of the singer helps, though, as it's useful for them to be able to adjust their own 'phones level.

Depending on the style of music you're producing, you may be able to get away with recording vocals at the opposite side of the room to the computer, to minimise fan noise, but ideally you'll want to put the vocalist in a different room altogether — unless you have a very quiet computer. The recording room needn't have a soundproof door, as you don't need very much attenuation to cut out computer fan noise — and in the absence of a talkback system, it's handy to be able to shout to the singer without having to open the door! In my situation, it's often enough to have the singer in the next room to the studio with the door open, provided that they're slightly around the corner, to take them out of the direct path of any computer noise passing through the door.

When recording in this way it's important to work with the control‑room monitors turned down fairly low, otherwise they'll leak into the vocal mic. It's also a good idea to turn off any metronome beeps, as these have a habit of leaking from the headphones into the vocal mic.

Most singers perform better with a bit of vocal reverb in the foldback, but not all software packages allow you to monitor with plug‑in effects. If you can, all well and good, but if not, you might have to resort to using a hardware effects box. Those using stereo‑out soundcards have two alternatives: either add a little reverb to the whole stereo mix during recording, or pan all the vocals to one side of the soundcard output and all the backing track to the other, then add reverb only to the vocal output. It's really a matter of picking what the singer is happiest with, but when you don't have all the facilities of a big hardware mixer to help, you have to improvise. Of course, adding hardware effects to selected parts is not a problem if you have a multi‑output soundcard and an external mixer.

Make Do & Mend

The ProTools screen, above, shows a four‑bar loop. Loop Recording has been enabled (see green highlighted Record button in transport bar). The different takes recorded after multiple passes are shown in the Audio Regions list on the right, and can be selected from a pop‑up list.

Unless you're working with an impossibly good singer, the need to redo or replace some parts of the vocal will inevitably arise. Punching in and out on a computer system is certainly possible, but many people prefer to create short 'patch‑up' recordings in a mode that doesn't overwrite the original audio. In most software packages, if you record a part in a track that already contains audio, you'll simply end up with a new section of audio sitting on top of the old one, and whichever is on top will play. This system allows you to easily decide whether the newly‑recorded sections are satisfactory or not. If you decide to keep them, you can then splice them into the main vocal track. Bear in mind that if your system does not allow monitoring of effects while recording there may be an unsettling change in the sound at the punch‑in point, as the original recording (with whatever plug‑in effects you've added) will be replaced by either a dry signal or one to which hardware reverb has been added. Most singers can deal with this, but you need to warn them first, or they'll think something has gone wrong.

Mixing Vocals

Antares' Auto‑Tune pitch‑correction plug‑in doesn't have to be used for a deliberately 'processed' sound — it also subtly corrects small pitch errors and can create useful double‑tracking effects.

Before getting too far into setting up a mix, I find it useful to check the audio and trim out noises, such as coughs, lyric‑sheet page‑turns, and so on, that occur during pauses in the vocal. If you can still hear some computer fan noise on the vocals, as you might if you had to record in the same room as the computer, you can either manually silence the pauses, which is very tedious, or use a 'strip silence' type of function to chop the track up into sections by removing silent pauses. This process works on a threshold system, rather like a noise gate, but if you find it's taken out something that should have still been there, you can usually extend the offending audio part manually to get it back. That's the wonder of non‑destructive editing — it may have disappeared from the screen, but it hasn't really gone! Alternatively, if the noise is fairly consistent and not too loud, simply insert a gate plug‑in and deal with it as you would when using hardware. As a rule, gating should come before compression, as that will make it easier to set the gate threshold, and always before reverb or delay plug‑ins, to avoid cutting off reverb or delay tails. Indeed, reverb can help disguise an edit or gate action that might not stand scrutiny in isolation.

Though you may have compressed your vocals during recording, you may still need to add a little more compression using a plug‑in. If this isn't enough to tame all the vocal highs and lows, you can always use the sequencer's mix automation to nudge the vocal‑channel level up or down as appropriate. Furthermore, because VST II plug‑ins can be automated, you may find that just increasing the compression ratio in difficult‑to‑tame sections does the trick.

The best reverb plug‑in for vocals won't necessarily be the most processor‑hungry one you have. Often a less dense reverb will sound more flattering, supporting the sound without clouding it too much. Adding some pre‑delay (between 50 and 100mS) will produce a subtle doubling effect that helps beef up weaker voices.

Whatever you think of the ubiquitous Auto‑Tune plug‑in, it's true to say that even good singers can benefit from a bit of subtle Auto‑Tune assistance. For singers who are already reasonably precise, I find that slowing the response speed a little from the factory default setting works best and provides the best trade‑off between pitch correction and unwanted side‑effects. Set the appropriate scale for the song, rather than relying on chromatic mode, especially if the singer uses strong vibrato — it can translate into unwanted semitone trills!

If there's a key change in the song, I've found the easiest way to handle it is to create a new vocal track for the key‑changed part, move the relevant audio into it, then open a new instance of Auto‑Tune, set to the new key. Occasionally you'll find a phrase that doesn't respond well to Auto‑Tune, either because it uses very dramatic pitch slides or because it includes notes not in the main scale. Again, this is easily dealt with by creating yet another vocal and moving the 'phrases to be left alone' into it. You don't have to duplicate all your plug‑ins for each track, as the reverb can be set up on a post‑fade send, and if you use a lot of insert‑type effects or processes you can route the channels via a buss/group, and then put the plug‑ins in the buss or group insert point.

Auto‑Tune is also useful for creating instant double‑tracking effects. Simply run a copy vocal track that hasn't been processed with Auto‑Tune alongside one that has. Introducing a small delay offset between the two and panning them to different sides helps the illusion. Using a short, tempo‑related delay on one or both tracks also produces a nice doubling effect, but unless applied very subtly it can be rather obvious, so make sure it suits the song.

A more natural way of doubling choruses is to copy a chorus from one part of the song and layer it with a chorus elsewhere in the song. The original performance will be slightly different each time the chorus is performed, so layering sounds in this way is bound to sound more natural. Avoid using Auto‑Tune, if possible, when you're doing this, or use it only on one part. Otherwise you'll tune out the small natural differences that create the double‑tracked effect. It's also best to save double‑tracking for parts of the song that need beefing up, as it can lose its impact if used all the way through a song.

Vital Voices

As you can see, creating a high‑quality, produced sound doesn't necessarily mean using an expensive studio. And there are lots of other tricks you can try in addition to the ones already mentioned. For example, use a tube emulation plug‑in to get more of a valve‑mic sound, or process a vocal part through a slow rotary speaker effect for that trippy, John Lennon sound. The main thing is to take care to get a good clean sound at source and to get that sound into your computer without degrading it. Process only as much as you have to, and use effects for a reason, not just because you happen to have the plug‑ins. Where you have several plug‑ins that do the same job, such as compressors and reverbs, take the time to try them all and see which works best. You may find that the best sound comes from the least expensive plug‑ins, so always go by your ears, not by the price tag.