Recording vocals, in a generic sense, has been covered in Sound On Sound on numerous occasions, but this time we're going to look at the process of recording and processing vocals using Apple's Logic. The process starts with getting a microphone signal into Logic, which usually means connecting a suitable mixer or mic preamp to the line input of your audio interface, or using an audio interface that has a mic preamp built in. Of course the choice and positioning of the microphone is just as crucial as in any other recording situation (see the 'Getting A Good Recording' box for some pointers). If you're using a capacitor mic, then the preamp or mixer must have phantom power — unless you're using a tube mic that is.
You may have a rack compressor that always sounds absolutely fantastic with your voice, but my inclination would usually be to avoid any form of processing while recording. In the bad old days, we needed to compress while recording to tape in order to maintain a high average recording level that would mask the tape noise, but with 24-bit recording you can easily leave 12dB or more recording headroom and not lose any quality at all. If you feel you must compress while recording, then at least make sure that you under-compress slightly rather than over-compress, as you can always add a bit more compression when you come to mix. The same applies to EQ — it's much easier to add it than to undo the effect of inappropriate EQ, so it's safest to record everything flat and then EQ while mixing if you need to. If your mic preamp has a low-cut switch, it may be worth using that to exclude very low-frequency sounds, such as passing traffic, but try a test recording with and without the filter to ensure that it doesn't compromise the low end of the voice in an unacceptable way.
The next issue with any computer recording system is to see whether you can get the latency value low enough for the singer to feel comfortable when monitoring the output from your audio interface. Nobody can give their best performance when what's coming back through the headphones is delayed to a noticeable degree. Most reasonably modern computer audio systems are happy to work with buffer sizes of 128 samples or lower, Logic included, but you may need to increase the buffer size if the computer is being forced to work particularly hard. Software instruments place the greatest drain on CPU resources and Logic's Freeze function can help you here by rendering these instrument tracks as temporary audio files to free up CPU power.
As a rule, I record with a buffer size of 128, which amounts to a latency of around four milliseconds (by the time you've added in the delay attributable to the A-D and D-A converters) at 44.1kHz and 48kHz sample rates — roughly the time it takes sound to travel four feet. Most singers don't perceive this delay, but please be aware that if you are using any DSP-assisted plug-ins such as those hosted by the Universal Audio UAD1 card or TC Electronic's Powercore, the latency will double and is then likely to be noticed. You'll also suffer greater monitoring latency if any of the tracks in your existing song use any plug-ins that themselves generate large delays, as Logic's plug-in delay compensation will (if active) delay all the other tracks to match. Most of Logic's processing plug-ins introduce only very short delays, but there are exceptions such as the new linear-phase equaliser. Note that plug-in delay compensation seems to be switched off when you first configure Logic, in which case you'll need to turn it on in Logic's Audio Preferences.
Where high latencies can't be avoided, you can still get around the delayed monitoring problem by taking a second feed from your preamp. Many preamps now have both jack and XLR outputs so you can record from one and monitor from the other, but a splitter lead can do the job just as well. The extra preamp output can then be mixed with the output from your soundcard, either in a separate monitoring mixer or in a master monitor controller that offers the facility to mix two or more inputs. You'll need to turn off Logic's software monitoring in the Preferences to avoid hearing both the direct and delayed sound, and you won't be able to add any plug-in reverb to make the singer more comfortable. However, if you have an old hardware reverb unit, you can patch this into the mixer you're using for monitoring and set up some 'comfort' reverb that way. You can feed the singer's phones directly from the monitor mixer's own headphone output or use the mixer to drive a separate headphone amplifier.
Getting A Good Recording
Firstly, hang up some acoustically absorbent material, such as a duvet or some heavy blankets, behind and to either side of the singer. This will help prevent sound bouncing from the room surfaces back into the microphone. Then set up the mic well away from walls, but avoid putting it exactly halfway between any two sets of walls, as this may exaggerate any room coloration — a couple of feet from the dead centre of a room is generally fine.
Always use a mesh pop screen between the singer and the mic, and leave at least a couple of inches between this and the mic — it won't work properly otherwise. As far as singer-to-mic distance is concerned, start with a gap of around six inches, as this is far enough away to avoid the worst of the proximity bass-boost effect, but close enough to give a good ratio of direct sound to room reflections.
Finally, provide the singer with a comfortable headphone mix, and don't record in the same room as your computer setup unless you really have to, otherwise you'll probably pick up some fan and drive noise. If you have no alternative, face the mic away from the computer and try to keep as large a distance as is practical between the mic and the computer.
If all has gone well, you should now have a clean recording of what was hopefully a good vocal performance. To turn this good recording into a great one, you can use Logic's Expander or Gate to remove low-level noise between phrases, you can use Compressor to add density and to even out the vocal level, and you can add EQ to fine-tune the tonality of the sound. The final touch is to add reverb, but if the vocal pitching needs a little work, there's always Logic Pro 7's new Pitch Correction plug-in to help you out.
As a matter of course, I usually get the singer to do at least two complete takes of each part, as well as revisiting sections that might need patching up. The complete takes can be handled on separate tracks, muting the first before you record the second. If you're lucky, you'll be able to use the scissors tool to compile the best sections from each to produce a definitive take, but more often than not you'll need to retake specific sections. The easiest way to do this is to create a new track that has the same Track Instrument as the original vocal take (or compiled take) and then use the automatic punch-in/out function to handle the recording for you.
First drag (from left to right) in the upper part of the Time Ruler bar in the Arrange window to mark out the section you want to record, activate the punch-in/out transport button and enable the record switch on your new track. Pressing Record will then start the process, playing from the current cursor position. Logic will play what's on the existing track right up until the punch-in point whereupon it will monitor the new recording source (the voice) and record it. When the system punches out, you'll hear the original track again. If the new section was OK, you can snip a hole in the original vocal track and then slide the new take into its place, holding down the shift key to prevent the start time of the insert from being accidentally moved. Once in place, you can use the crossfade tool to avoid glitching if you've cut in the middle of a phrase, but most times it's best simply to arrange the edit (punch in and out) points in the pauses between phrases.
During this compiling and optimising process, you may also find that some chorus sections or other repeated phrases sound better than others, in which case you may wish to copy them and replace the less good ones. Using Logic's new Marquee tool, you can identify your source section without having first to cut the sequence into pieces using the Scissors tool and then copy it using the usual Alt-drag procedure after first switching back to the normal Pointer tool.
Once you're happy with the quality of the vocal part, you can start to process it. Logic's Gate is perfectly adequate for removing low-level noise between words and phrases, and its side-chain filters are useful if there's any other very low- or very high-frequency sound present that might otherwise trigger the gate and open it when it should be closed. The gate should be the first process used, so it's best to put it in the top Track object insert slot. If you set the side-chain filters to 150Hz and 6kHz respectively, the gate should trigger reliably on vocals while ignoring things like traffic rumble or fan whine. For vocals, it's usually best to set the gate to open as fast as possible, but to close over a period of around a quarter of a second, so that naturally decaying notes aren't cut off too abruptly. You can see if this can be improved upon by trying faster release times, as a lot depends on the performance and the level of noise present.
While gates can apply almost infinite attenuation when closed, you may get a more natural sound by setting the gate range to around 12dB, which should still be enough to hide any noise (unless your recording is really noisy). I've tried Logic's own noise-reduction plug-in, but found it rather disappointing, and though an Expander will work just as well as a gate in this application, there's no real advantage under normal conditions.
Once you've gated the vocal track to clean it up, you can then compress it by inserting Logic's compressor in the next available insert slot. For vocals, I tend to use the RMS compressor mode with the Knee slider over towards the Hard end of its range. The ratio can be somewhere between 4:1 and 8:1, depending on how assertive the compression needs to be. A fast attack followed by a 100-200ms release time usually works well, but if you like your compressors to audibly pump, then try a shorter release time. That leaves the Threshold control, the setting of which depends on the level and dynamics of the vocal recording. A good rule of thumb is to adjust it until you get around 6dB of gain reduction on the loudest peaks if you are being subtle, or up to 16dB or so if you're after an obviously compressed rock vocal sound. However, be aware that for every decibel of noise reduction applied, any background noise and room ambience on the recording will come up by exactly the same amount between phrases, which underlines the need to capture a clean recording in the first place. That's another good reason to gate the track, even though it may not appear noisy prior to adding compression.
Logic's Pitch Correction plug-in can be placed before or after a compressor and, intuitively, I feel it has a better chance of working properly if it comes afterwards, as it will receive a more constant level. Having said that, this plug-in is so good that it copes with all normal vocals, compressed or not. Normally I only use pitch-correction where I really have to, and you can slow the rate of pitch-correction right down between the phrases that need treatment, using Logic's automation, which has the effect of reducing the amount of processing applied — only long, sustained notes are then corrected. You can also automate the plug-in bypass button if you prefer to do it that way.
An alternative approach is to split the audio part over two tracks with identical plug-ins and track settings, then insert the Pitch Correction plug-in on one of them. By moving just the parts that need correcting onto the track with the Pitch Correction plug-ins, you can easily control where the processing is and isn't applied without having to use any automation. If you've yet to upgrade to Logic v7.1 then it'll be necessary to duplicate the track plug-ins (which can be copied over, complete with their settings, using the Audio Configuration window) rather than routing both audio tracks to a Buss object and putting the common plug-ins in its insert points — although this saves on processing power, the lack of plug-in delay compensation in Logic's busses will cause the vocal part to be delayed slightly. You can fix this manually using a negative track delay if you're really short of processing power, but under normal circumstances, duplicating a gate, a compressor, and possibly an equaliser won't add significantly to the processing load.
As with most pitch-correction plug-ins, the most natural results are obtained by setting the response time as slow as possible commensurate with still controlling any obvious pitch wandering. If you set it too fast, natural inflections start to get ironed out, and the result becomes somewhat robotic. Although this misuse of pitch-correction has been successful as a special effect on a number of pop records, it's not the way to go if you want a natural-sounding vocal performance.
EQ is best applied after pitch-correction, and what you need to do, as always, is emphasise the positive aspects of the source sound while reducing any unpleasant characteristics. Often a sizzle is added to the high end by applying a broad boost (around two octaves wide) at anywhere from 8kHz to 12kHz using Logic's Channel EQ (the standard version, not the new linear-phase option). It's also worth setting up a fairly narrow band (half an octave or so) and then sweeping this through the mid-range to see if it picks out any nasal or boxy artefacts that need suppressing. More experienced users will be able to estimate the problem frequency and then try the cut without having to sweep first, which can help maintain a better sense of perspective. Which approach you prefer is up to you — I still tend to be a bit of a sweeper! EQ seems to be a very personal thing, where some people use hardly any and others pile it on. Different musical styles demand different approaches, but if the end result sounds right, then it is right. However, never judge the result of EQ in isolation — it's essential that you hear the part in the context of the complete mix.
Until very recently, the majority of native-powered software reverbs have been, frankly, rather disappointing. Logic's Platinumverb used to be top of the Logic reverb range, but it's really not that great. It is fine for demos, and it suits some instruments, but it's not a classy vocal reverb by any stretch of the imagination. As for the Gold, Silver, Lead, and Cardboard variants, just forget them. If you're using a version of Logic that only offers these, and you don't have any third-party alternatives, then the best option is to use a mid-price hardware reverb unit and feed it from one of Logic's aux sends via a spare output on your soundcard. If you only have stereo digital and analogue I/O, then choose a reverb unit with a choice of digital I/O so you can feed it directly from the soundcard's S/PDIF output. Its output, analogue or digital (depending again on what your soundcard offers), can then go back into two soundcard channels routed to a stereo Input or Aux object within Logic, which will then act as an aux return.
Having said all that, anyone with Logic Pro 7 also has Space Designer, a fantastic reverb, at their disposal. It is a bit processor-intensive, as it uses impulse responses and convolution techniques, but you can halve its appetite by choosing a halved Sample Rate setting at the left of the plug-in window and then clicking the Preserve Length box. This in effect halves the frequency response of the reverb, giving it an upper limit of around 10kHz, but that's still enough to emulate most classic reverb effects.
Normally Space Designer would be placed in a Buss object and fed from track sends — even if you don't have plug-in delay compensation, who cares if your reverb comes back a few milliseconds late? It's the same as increasing the pre-delay very slightly. Just remember to ensure the Space Designer mix controls are set so that the Direct signal is off and the Reverb signal is at maximum. You can then adjust the amount of added reverb using the send controls, and in most cases somewhere around 20 percent reverb sounds about right.
As to which reverb setting to use, you're spoilt for choice, especially if you have Logic Pro 7 with all its extra reverb impulse responses. There are many vocal plates and vocal studio presets to try, but my advice would be also to try out some of the less likely-sounding presets such as the woods and forests, as some of these sound stunning on vocals. Even the Station Platform preset can sound great. Whereas the synthetic reverbs in Logic sound grainy and don't really gel with the original sound, Space Designer behaves just like a high-end studio reverb, becoming part of the sound itself rather than seeming to be layered on. For conventional pop music, the impulse responses based on other hardware reverbs rather than on real spaces often work best, as that's what we're used to hearing.
Some people find the need to de-ess vocals, and Logic has a plug-in for that too, though I don't think it works as well as a dedicated de-esser box, such as the one made by SPL. In most cases, changing the mic position or choosing a different mic works in all but the most extreme cases. In fact, I find that the most common requirement is to make the vocals sound more crisp, which can be done using 'air' EQ in the 12kHz range or by using Logic's Exciter plug-in, which essentially resynthesizes the upper harmonics based on what's going on in the mid-range. It's important not to set the filter frequency too low (1.5-2kHz seems good for vocals) and not to over-process, but switching to bypass on a regular basis should ensure you're not overdoing things.
Although there is a lot of esoteric hardware and software available for processing vocals, the plug-ins that come as part of Logic Pro are capable of doing a great job if used correctly. However, the most important thing is to start with a good, clean recording made in a well-damped or acoustically sympathetic space, as any boxiness caused by the room can't be fixed afterwards. After that you need to use your ears to judge what the sound needs, rather than processing just because you feel you have to. If you get the best sound without compression or EQ, then don't use it. Similarly, don't smother the sound in reverb just because you can. On many mixes, I find it's appropriate to use less compression and then compensate for more excessive level changes using Logic's level automation. The less you have to do to a vocal to get it sounding right, the more natural the end result is likely to be.