The Art Of Recording Electronic Instruments
Recording acoustic instruments is a fine art, but electronic instruments present their own set of problems. Craig Anderton explains how to do it properly.
While it's tempting to think that recording synths and samplers simply involves feeding a direct output into a mixer or recorder, there are problems unique to recording electronic instruments that must be considered if you want the best possible sound. Unlike acoustic instruments, where the main difficulties involve miking and acoustics, electronic instruments need no mics and include no inherent acoustic space. However, they do come with problems such as background noise, lack of expressiveness, a 'sterile' sound when played side-by-side with acoustic instruments, timing inconsistencies and, often, monophonic sound generation.
Addressing these problems while taking advantage of the many features unique to electronic instruments can add life and sparkle to your recordings, so what follows is a series of tips designed to make your synths and samplers come alive during the recording process.
Tape has a limited dynamic range, which can cause complications with certain types of synth patches. It's important to avoid strong signal peaks, which are bad enough with analogue recording (you'll get distortion) but can be disastrous with digital recording — go much over zero VU, and you'll hear a very nasty splattering type of distortion.
The first line of defence is proper synth programming. Detuned oscillators, though they sound nice and fat, create strong peaks when the chorused waveform peaks occur at the same time. This can raise the average level enough to saturate the tape. The solution is simple: drop one oscillator's level back to about 50-75% of normal. This will still allow for an animated sound, yet the peaks won't be as drastic and will be less likely to cause distortion.
High-resonance filter settings are also troublesome, since hitting a note at the filter's resonant frequency creates a radical peak. To print the maximum possible signal level on tape, use as little resonance as is necessary. Consider using a parametric on mixdown to add resonance rather than recording highly resonant sounds.
For those situations where you don't want to modify the patch sound, use a limiter to control dynamics, set to fast attack and moderate decay. Since you're mostly using the limiter to trap short peaks and transients, set the threshold fairly high, and use hard knee limiting. This will leave most of the signal relatively unaffected, but peaks will be damped to a safe, non-distorting level.
Remember also that limiting and compression can be used as effects. It seems most bass parts are limited to some degree or another, and limiting synth bass parts can help 'smooth out' the sound and add more punch (as well as give a more consistent level). Sounds with chorusing or other processing that causes significant level variations can also benefit from limiting. However, be careful when limiting an equalised signal. The limiting will partially undo the effects of the EQ, which leads you to turn up the EQ, which creates more peaks, so you add more limiting... It's usually better to add EQ post-limiter rather than pre-limiter.
Drum machine sounds can also benefit from limiting. We've become accustomed to hearing limited drum sounds over the years, so adding a bit of limiting to a drum machine can create a more 'commercial' sound. (In fact, drum samples are often recorded with some degree of limiting). However, limiting an entire kit can cause undesirable side effects, such as pumping and breathing. To circumvent this, split the drums into two submixes, with the kick, snare and toms feeding a limiter, and the assorted percussion and cymbals feeding a non-limited bus. The main drum sounds will have more punch, but the lighter, more accent-oriented sounds will retain their original dynamic range and not be subject to the side effects of limiting.
In the days when analogue tape was king, many engineers used it (knowingly or unknowingly) to perform soft limiting and generate some distortion on drum sounds by recording well into the red (overload) zone of their VU meters. Unlike digital recording, where the onset of distortion is very abrupt, with analogue recording distortion increases as a function of level. Turning up the record level to the point of mild distortion can 'crunch' the drum sounds in a way that is quite pleasing, especially if it is selectively applied to different drums (a snare seems to work best, but you be the judge).
By the way, this is also a useful sampling technique: record a drum on tape so that it crunches, then sample the taped sound.
The low end of the dynamic range, where noise occurs, is also important. Even today's digital wonder boxes generate some noise, and it's important to keep this to a minimum.
Always set the master volume control on your instrument to the maximum possible level. In most devices this control affects the digital (rather than the analogue) stages and must be set at full to get the maximum range out of the unit's digital-to-analogue convertor(s). If distortion occurs with the master volume up full, try lowering the level elsewhere in the synth (perhaps there's a page with an 'output level' parameter) rather than altering the master volume.
Speaking of internal level-setting parameters, to ensure the best signal-to-noise ratio these should also be set to the maximum possible level short of distortion. Most synths have several level adjustments: settings for individual oscillators; the envelope levels controlling DCAs; final output mixer; onboard signal processing trims. These must be tweaked with the same care with which you set the preamp, submaster, master, etc. controls on your mixer.
Older synthesizers often generate some constant background hiss. While standard noise gates will help take care of this, I've found that downward expansion devices (like the Rocktron Hush) do a more natural-sounding job of eliminating noise. If you have neither form of noise reduction handy, then try the pre-emphasis/de-emphasis trick: boost the highs at the synth (usually by changing to a waveform with more harmonics and/or increasing the low pass filter cutoff frequency or envelope amount) and use your mixer's equalisers to roll off a bit of the top end to compensate.
With samples, it's often possible to use the sampler's on-board filtering options to reduce noise. The trick is to envelope the filter so that the maximum high frequency response occurs during the attack, and then during the sustain or decay the filter closes down more to reduce noise. This can really help clean up otherwise marginal samples.
When recording parts that consist of multiple instruments playing single note lines (eg. brass sections, some pads, vocal choirs, etc.), try overdubbing each individual line instead of playing chords. Although this uses more tracks (but remember that you can always bounce them down to free more tracks) and takes more time to record, you can articulate each line with respect to vibrato, dynamics, etc. You'll be amazed at how much realism this technique can add to recorded parts. To avoid confusing your new part with previously-recorded takes as you play, monitor in stereo with the older parts panned left or right, and the new part in the centre.
One problem with synthesizers is that it is difficult to add subtle articulations to each note as you play. For example, a wind player will vary embouchure, dynamics, and phrasing to add interest, whereas a synthesist playing a wind part will often just press the key down to start the note, and hold the note for the desired amount of sustain. This makes for a boring, static sound.
The solution is to use real-time controllers (footpedal, mod wheel, pressure, etc.) as you play, but it's difficult to work these with great precision and play a part at the same time. Instead, record the part into a sequencer that's synced to tape, and edit the sequence to introduce slight variations — an occasional bend up leading into a note, some controller 7 (master volume) data to vary the dynamics of individual notes instead of just overall volume (eg., have occasional notes fade in and fade out over time), 'swell' the filter using a pedal or aftertouch, and so on. Now run the sequence back while synced to tape, and record the edited synth part on tape (unless you're using virtual tracks, in which case you don't need to record the sequenced part on tape, just play it back in parallel with the tape tracks).
When playing live, you're often constrained by the amount of equipment you can take with you. In the studio, you can construct sounds using not just your own gear, but any synths the studio happens to have available. Various forms of synthesis have different strengths, so layer several synths via MIDI and use each to provide a particular part of the overall composite timbre.
For example, samplers can sound very realistic, but lack variation because a sample is essentially static and the attack and sustain characteristics are pretty much constant. An FM synth may not offer the same 'photographic' level of realism, but it can produce wide timbral variations — particularly on a sound's attack — that can be modulated by velocity. I've used this to good advantage on harp and plucked string sounds, where the FM synth provides the pluck and the sampler the body of the sound. Grafting the two elements together produces a far more satisfying effect than either one by itself.
One minor problem is that if the two sounds sustain for any length of time, the timbral difference may become too noticeable. Therefore, you might want to set a fairly short decay on the attack sound and a bit of an attack rise on the sustain sound so that it doesn't overwhelm the attack component.
Sometimes you can achieve great results fairly painlessly by pulling up like-named patches (eg. vibes, nylon guitar, etc.) on different synths, layering them, and recording the results. This technique also works well with drum machines if you assign different snares, toms, kick, etc. to the same MIDI notes. For example, the original Alesis HR16 sounds a little bottom-heavy compared to some of the brighter, more contemporary sounds on the SR16. I send the same notes to both, and bring up the HR16 in the mix just enough to add some more depth to the SR16 sounds.
Some synths respond sluggishly to MIDI data, so while you may be hitting the keys at just the right time while doing an overdub into a multitrack recorder, on mixdown the recorded result may sound 'late' compared to other instruments. You may also want to vary timing for aesthetic reasons, perhaps delaying a drum part for a more 'laid-back' sound or advancing a hi-hat part to 'push' the song. Delaying a part after recording is easy — just feed the track through a delay line — but how do you advance the timing?
If you have an open tape track, just flip the reels so that your tracks play backward, then use a delay line to delay the track you want to advance. Record the delayed part into an open track, and when you re-flip the tape for normal playback, the track will be advanced by the amount of delay you added. This technique is particularly useful with string parts, whose slow attack envelopes often create a feeling of 'lateness'.
Although you can always create stereo effects through ambience generators such as reverbs, sometimes only a true stereo instrument will do. There are several ways to create such an effect.
Some samplers allow for stereo recording, but even mono samplers can create stereo effects. For example, with the EPS it's possible to assign two sounds to different layers, and pan one to the left and one right. But what if you only have a mono source and want to convert it to stereo?
Although detuning two layers panned hard left and hard right is one possibility, this may produce unpredictable results if the sample is played back over a mono system. A better solution is to copy a mono signal to two layers, pan them left and right, then change the start point of one of the samples. The further you place the start point into the sample, the greater the stereo effect (until you reach a point of diminishing returns, after which further start point movement doesn't affect the sound very much). If you can modulate the start point to create a bit more variation, and this doesn't induce glitching, so much the better.
Small start point changes may produce out of phase effects if the signal is played back in mono, so check carefully for this. If necessary, move the start point further into the sample. It may be necessary to slightly lengthen the attack time if changing the start point creates an overly-abrupt attack.
This technique is particularly useful with pads and other sounds that have non-percussive attacks. If you move the sample point further back in a percussive sound, you may lose the initial transient.
Yamaha's later-model DX and TX line of instruments have a voice architecture where different individual programs can be arranged in what's called a Performance. In this multi-timbral mode, up to eight different programs (called 'instruments') can be set to specific MIDI channels, cover particular note ranges, and (in conjunction with individual outputs on devices like the TX802) be panned to anywhere in the stereo field. Even if the synth has only stereo outputs, each instrument can usually be at least panned to left, right, or centre.
This allows for some spectacular stereo effects if you use the same sound program for each part of the Performance, but restrict the programs to particular note ranges then pan them to various places in the stereo field. For example, suppose you have a lead part that covers a range from C1 to C5. Set the first instrument to only play notes C1-F1, the second to play notes F#1-B1, the third notes C2-F2, etc. until the eighth instrument plays F#4-C5. Now pan the instruments across the stereo field as shown in Figure 1. As the part plays back, lower notes will appear on the left, and higher notes on the right.
Other synths have similar capabilities. For example, the Ensoniq EPS 16 Plus has eight layers whose note ranges can be restricted and can be panned anywhere in the stereo field. The Peavey DPM3 can assemble four program links into a Combi; note ranges can be restricted, but panning has to occur within the programs that make up the Combi (Figure 2).
One of the problems with synthesizers is that they sound out of place in a track with lots of miked sounds. Sounds recorded through a mic will always include some degree of room ambience, even though in many cases the sounds are recorded in fairly 'dead' rooms. One answer is to put the miked and direct sounds through a common hall reverb or similar effect during mixdown, but this still may not sound quite right. Try this: during recording, patch the synthesizer sound through a reverb unit that simulates the sound of a small, fairly dead room. This should be just enough to give the synthesized sound a bit of acoustic depth. Now when you put the synth sound through the main hall reverb during mixdown, it should fit in better with the other tracks.
Don't forget that signal processing is often a vital part of a guitar's sound. Common choices are chorusing, discrete echoes for solos, distortion, and wah-wah (filter sweep). Feeding the synth or sampler through a device like a Scholz R&D Rockman or (better yet!) a miked vintage tube guitar amp could give just the sound you seek. Other options are tube preamps (which really do warm up synth sounds) and the 'speaker simulator' modules available for guitar, as well as multi-effects signal processors that provide this function.
When recording synthesized piano, organ, and other 'real' sounds via direct injection, try mixing in a bit of miked sound of you playing the keys. This should be placed subtly in the background — not loud enough to be obvious, but just noticeable enough to give a subtle aural cue. The natural quality the sound acquires may surprise you.
There's only so much a keyboard workstation can do, and if there's onboard signal processing, odds are it will be less sophisticated than dedicated rack mount units. As a result, you'll generally have a higher quality sound if you turn off the synth's internal effects and use outboard effects instead.
Another consideration involves polyphony. Many synthesizers layer two or more notes on one key to thicken up a sound (perhaps detuning layers slightly to produce a chorus effect), but this reduces the available polyphony. Use an external delay to produce chorusing instead, and you'll be able to play more notes without having them cut off.
Better yet, if you have the tracks, overdub a second part to produce chorusing instead of doing it electronically. The part will sound more 'humanised'.
Speaking of overdubs, if you want to double a part, change the deck's speed a little (and retune the synth to compensate) prior to recording. The pitch shift will produce a slight timbral change that will add more variety to the overall sound.
The LFO panning option found on many synthesizers is useful as far as it goes, but can produce an overly regular, boring sound. However, with short percussive instruments (eg. claves, tambourine hits, cowbell, etc.) that don't sound too often, panning can work very well to add animation. Because the sound is short, you don't hear it pan as such; instead, each time the sound appears, it will be in a different place in the stereo field. If you have a rock-solid kick and snare, a percussive part dancing around the stereo field can add interest.
To create more interesting panning patterns than most LFOs can provide, and assuming you have envelopes with multiple break points, take a mono sound, copy it to two layers, then pan these layers left and right. Adjust the envelopes for opposite settings so that as a signal increases in one channel, it decreases in the other (many instruments let you invert an envelope's modulation so you can simply add negative envelope modulation, thus obviating the need to reprogramme the various level parameters).
It has always been acknowledged that recording acoustic instruments is a fine art, but the recording of electronic instruments requires equal care and dedication to achieve suitable results. I hope the above tips will help you create better-sounding recordings, where synthesizers and samplers blend into a track and form a unified, cohesive listening experience. As always, the most important consideration is to use your ears. Try to be the most critical listener in the world as you record and mix, and don't be satisfied until you get exactly the sound you want; if you have doubts about a sound, then work on it until you're satisfied. Good luck!
Craig Anderton is the author of 11 books including MIDI For Musicians and Home Recording For Musicians. He is also a recording artist who has played on, mixed, or produced, ten albums including his latest CD, Forward Motion.
Feature by Craig Anderton
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!