DUB STAR

Basic Overdubbing

Published in SOS March 2001
Printer-friendly version

Technique : Recording/Mixing

Hugh Robjohns and Mike Senior explain how to get great results when creating arrangements one track at a time.

Fifty years ago, the recording of pop music was a very different activity. The recording equipment of the time was so crude by modern standards that a complete performance had to be captured in one go — you simply had to keep doing takes until everyone got it right. While this method had its benefits — often those recordings had a real excitement and sense of performance — it did make great demands on the musicians involved, and it was not suitable for the multi-instrumentalist working alone.

Modern multitrack technology allows an entirely different approach to recording, because each instrument can be recorded onto a new track while listening back to those already recorded — a process known as overdubbing. What's more, you can overdub only sections of each track, in order to improve on the performance or correct mistakes.

In order to build up complete compositions using overdubbing, you need a multitrack recorder which is designed to allow new tracks to be recorded independently while the original tracks are replayed in sync. In addition, you need to be able to process microphone signals and other sound sources on their way to the recorder, as well as being able to create monitor mixes from the multitrack recorder's outputs, both for you and for your performer.

Traditionally, a mixer would be used for both input conditioning and for monitoring purposes — input channels preamplify and condition sound sources, routing them to group busses connected to the multitrack inputs, while monitor channels accept the line-level multitrack outputs and provide controls for setting up the required number of independent monitor mixes. Note that the controls for these two different signal paths often occupy the same physical space in the complex 'in-line' mixers that you can often see in pictures of professional studios. Mixers using an in-line configuration can be very flexible and can save space in large installations, though they are generally more expensive and difficult to operate than consoles where input and monitor signal paths are kept completely separate.

Having a suitably sized multi-channel, multi-buss mixer is vital where you're recording a number of musicians simultaneously. However, many home recordists never require such facilities because they build their music up one track at a time. It is because of this that so many voice-channel processors have appeared on the market, providing high-quality preamplification and processing — such as compression, expansion, equalisation, de-essing or enhancement — without breaking the bank. However, multi-channel monitor mixing is still required by the home studio musician, though such a monitor mixer can often be much cheaper because it requires fewer facilities than a full-scale recording mixer.

Though musicians who begin their tracks using a sequencer and MIDI instruments are unlikely to encounter any problems changing these parts and overdubbing individual audio parts over them, anyone doing more traditional band recording will have to prepare for overdubbing right from the start of the recording process if it is to be successful.

When recording a band, it is common practice to begin the process by setting everyone up in the studio and recording them all performing simultaneously, in order to get the feel of a live performance. However, if you want to be able to repair or replace sections of individual tracks later then you need to have excellent separation between the individual instruments. If you can hear any instrument clearly on any recorded track other than its own, repairing any of its mistakes using overdubbing will prove very difficult, if not impossible. By way of example, if the lead guitar amp is picked up on other mics in the studio (drum overheads, perhaps) then it will be impossible to replace a bad solo by overdubbing — the original solo will still be audible as spill on the drum overheads and will conflict with any overdubbed part. As a general rule, spill is the major enemy of overdubbing.

Once you have a basic take, or a MIDI backing track, which sets the structure of your song, you can set things up to start the overdubbing process. The already-recorded tracks ought to be made safe, so that they aren't recorded over by accident, while the track to be overdubbed will need to be armed for recording. The replaying tracks from the multitrack recorder should be routed through your monitor mixer to the main stereo mix output to provide a monitor balance in the control room. What's more, arrangements for supplying an independent cue mix (the monitoring feed which the performer hears as they overdub) should be made if required. If you are recording someone else's performance, it is worth sorting out all the routing and the balancing of the control-room monitor mix before you get the performer into the studio — nothing takes the wind out of a musician's sails like having to wait around while these tasks are carried out.

The key to good overdubbing is setting up a good cue mix. The performer needs a suitable balance of any tracks already recorded as well as a feed from their own instrument or voice. If a performer is being recorded through a microphone, then it will probably be best to provide them with the cue mix through a pair of headphones, in order to avoid any spill from this mix being picked up by the microphone. However, if you're recording DI'd electric guitars (or if your guitarist has a long lead to feed a cab being recorded in another room) it can often be more pleasant for him or her to do the recording from the control room, playing along to a mix on your main monitors.

If you are recording yourself, or if the performer is going to be listening to the main monitors while overdubbing, you can simply set up the faders on your monitor channels as required — effectively making the main monitor mix and the cue mix the same. However, if the performer requires a different mix to that on your main monitors (for example, if the performer needs to hear a click track very loud in their headphones for timing, while you'd prefer to listen closely to what is being recorded) then you'll have to set the cue mix up using one of the auxiliary send busses on your monitor mixer.

If you decide that you need to have a separate cue mix, you should choose a prefade send if at all possible. If you are unable to do so, then you'll have to be careful when adjusting the control-room monitor mix, because any changes to the fader levels will also be reflected in the headphones of the performer. Given how crucial the cue mix can be to a good performance, it would be best in this case to avoid adjusting the fader levels at all once you've successfully balanced the cue mix. While the feeds from recorded backing tracks and any MIDI-controlled sound sources will be sent from the monitor mixer channels handling these outputs, we haven't yet catered for the performer hearing himself.

Because overdubbing usually involves rerecording small portions of a given track (also referred to as 'dropping in' or 'punching in'), the performer needs to hear their originally recorded version before and after the section which is being rerecorded, in order to match it. Furthermore, they need to hear the input signal to the recorder whilst actually recording, in order to hear what they're doing. This needn't be a problem if you have a multitrack recorder which automatically switches to monitoring the input of any record-enabled tracks whenever it drops into record mode: this allows you to build up the entire cue mix using only the monitor mixer's aux sends.

There are occasions where you might wish to feed the cue mix directly from the overdubbing musician's input channel. For a start, some musicians actually prefer to be able to hear themselves all the time, not just when actually recording — this means that they end up hearing two versions of their performance both before and after a drop-in. This can be useful, but can also take some getting used to.

Another common time when you might want to feed the performer's cue mix from their input channel is when working with computer-based systems which exhibit monitoring latency. If you monitor any signal live through such a system, the latency will cause the performer's sound to arrive a little late in the headphones — this can be very disconcerting, so it is best to monitor the input signal only. Because monitoring the input signal alone can make dropping in difficult, people working with such systems usually adopt a slightly different way of working to that traditionally used with stand-alone multitrack recorders: for example, rather than doing a small traditional drop-in to correct a mistake, it makes more sense in such systems to rerecord the whole line again on a new track, subsequently editing a correction in the place of the error. The erroneous track can be fed to the cue mix via the monitor mixer while recording on the second track if this is useful to the performer, or else can be muted.

If you're using a mixer only for monitoring purposes, it may not be easy to feed the cue mix from the input channel. In such cases, an alternative technique is not to include the source channel in the cue mix at all, and to persuade the performer to only wear one side of their headphones, allowing them to hear their instrument directly. This works well for most instrumental overdubs, though you may have to pan the cue mix to one side if you wish to stop the sound from the unused headphone spilling through and being picked up by the mic.

Sometimes, an effects process is integral to the way a musician plays or sings a part to be overdubbed, and the performance will interact with the effect being used. For example, a delay might be integral to the performance and will probably affect the speed of playing. Likewise, an amp simulation will often be vital to a guitarist's performance. In such cases it's always sensible to record the original sound and the effect's output to separate tracks simultaneously — this gives you the option of reprocessing the original sound when mixing, if necessary. It is for similar reasons that many engineers, when recording electric guitars, record not only the signal from the microphones on the guitarist's cab (or line out from their amp simulator), but also record a clean DI signal, that can later be 're-amped' and recorded again if the original sound was wrong.

In addition to effects that are recorded, it is always desirable to have independent control over the reverb level in the cue mix. This is particularly important when recording vocalists, as they often find it more satisfying to hear the sound of their voice with plenty of 'monitor reverb' applied, and therefore tend to give a more confident performance. In order to do this you will have to make sure that your reverb's output feeds into channels or dedicated returns that will allow you to send to the cue mix busses.

Having decided on the sort of cue mix you're going to have, and how you're going to route all the necessary elements to it, you then need to actually balance these sources. A good cue mix can make all the difference to a performance, while a bad one can make it hard to perform in time, in tune, or in the desired way. Because of this, many engineers bring a pair of headphones into the control room and allow the musician to balance it for themselves. However, this may not provide the best solution — what a performer wants and what they need can often be very different things...

Communication While Overdubbing     The final elements that may need to be added to the cue mix are communications signals, which allow the engineer and performer to speak to each other. Such signals, sometimes called 'talkback', may not be necessary for the way in which you're working — obviously, if you yourself are both the engineer and performer then there are unlikely to be communications problems, and there's also no need for talkback if the performer is in the same room as the engineer. On the other hand, if you have an electric guitarist recording in another room, then you might well need two microphones for this purpose: one each for engineer and performer. If your mixer has no built-in facility for this, then such mics can be fed directly into your monitor mixer and fed to the relevant mixes from there. Do bear in mind, though, that you should avoid routing the engineer's talkback mic to the main monitor mix, unless the talkback system incorporated in your mixer automatically mutes the monitors when this mic is active, because of the potential for feedback howlround. What's more, it's usually worth muting talkback mics while actually in record mode, to avoid distracting the performer with any unrelated sounds these mics might pick up.

If the performer in question has difficulties with tuning, there are a number of ways in which the cue mix can be tweaked to improve this. Firstly, the level of the cue feed from the performer can be increased or, alternatively, the performer can be encouraged to remove one ear of their headphones in order to be able to hear themselves more clearly at source. The overall headphone level can also have an effect on a performer's ability to stay in tune, so it is worth experimenting with this, within the realms of the musician's comfort, tolerance and safety. Monitor reverb can be turned down in level and/or the reverb time shortened, as reverb can sometimes obscure short-term tuning inaccuracies — bear in mind, though, that the unfamiliarity of this could actually make the singer feel more uncomfortable and thus prove counterproductive.

If there are any elements of the mix which are detuned or unpitched, there can be a good argument for reducing these in level if tuning is proving to be a problem, while there is an equal case for increasing the levels of any well-pitched harmonic elements, such as the bass and rhythm guitars or the keyboard pad. In some cases, it's worth programming an extra MIDI part, playing the desired line, to which the performer can tune, though this can sometimes be confusing if too similar a sound is chosen and it can also risk making the performance mechanical if the guide part is rigidly quantised. If all else fails, and the singer seems unable to get out of a rut of playing or singing certain notes out of tune then, if this is possible on your system, try varispeeding the whole performance (transposing any MIDI parts if necessary) by a semitone up or down — being in a slightly different register, and possibly having to play with different fingerings, can sometimes help performers break out of a bad tuning habit.

Timing problems can often be solved by simply increasing the levels of the more rhythmic elements, while backing off any tracks with solos or lead parts which pull or push the beat. However, whenever you change the cue mix to solve a tuning/timing problem, you should be aware that it will also change the way in which the performer interprets the line. For example, if you fade up the drums to improve timing, you may find that the manner in which the performance is stressed will also change.

Once the cue mix is in place, the session proper can begin. Different engineers and performers have different ways of going about this, but the overall aim of all these approaches is the same — to achieve the best interpretation while still retaining the required level of technical proficiency. Though views on what amounts to the 'best' interpretation and the 'required' level of technical proficiency may vary, the ability to balance the pursuit of these two often conflicting goals is the art of good overdubbing. If you allow yourself to get too involved with a succession of very detailed drop-ins, for example, then any technical improvements will often be made at the expense of a natural emotional flow throughout the entire section or song. Balance any periods of dropping in with periods where you go for whole takes — it is very easy to lose the plot a bit when overdubbing, so it can really be worth stepping back a bit from it every now and again.

Whenever you do a drop-in, it is essential that both engineer and musician agree on where the punch-in and punch-out points are to be. Mistaken drop-ins are not only frustrating for the performer, but may also need repairing, particularly if you are working without the benefit of an 'undo' function. In addition, you want to encourage the performer to play or sing as normal on either side of any drop-in they do, in order that the beginning and end of the section are replaced as seamlessly as possible.

If you need to keep redoing a drop-in, it can be worth making sure that each new drop-in preceeds the previous one, and each drop-out is later than the last one, in order that only one set of punches can possibly become audible. Because there is often little leeway in positioning a drop-in, it is usually better to rehearse the repair until you're sure a single overdub will do the trick. If your multitrack recorder can drop in automatically then you can rerecord a section multiple times without any of these problems. However, if your recorder exhibits monitoring latency, it may prove more successful to record the new section on a separate track and edit it in once you've got it right. In fact, no matter which recorder you're using, you may wish to do multiple complete takes and assemble a final version from these later — a process commonly referred to as 'comping' — either by bouncing sections to another track or by using the recorder's digital editing.

Learning how best to overdub can take a lot of practice, particularly because sensitivity to the unspoken requirements of the performer plays such a large role. However, if you're careful to follow the advice above, you should be well on the way to getting high-quality results every time.