Having dealt with electric instruments and effects, Brian O'Shaughnessy's guitar recording series this month goes 'unplugged', in search of richer pickings

In this article I'm going to look at recording acoustic guitars. But before we go into practical details, some general comments.

First, a cautionary tale. Once upon a time, a young recording engineer was very pleased with himself when he discovered how to get a great sound with his old acoustic. The next day, he was working with a band who wanted an acoustic on one of their songs; no problem, he thought, I'll just use the same guitar and recording set-up. But, try as he might, he couldn't get the thing to sound right.

After about an hour of trying everything in the studio, different mics, compressors and EQ, the band, who had been assured of a great sound in a few minutes, were starting to get distinctly annoyed and the engineer was looking worried. In desperation, he picked up the guitar himself and found it was out of tune. A quick tune-up later, and everything was fine. I know this story is true, for that young engineer was me. Sorry to labour a point, but the moral is clear: always listen to the source sound before reaching for electronic solutions.

When recording acoustic guitars, the quality and condition of the guitar is even more important than it is with electrics. Make sure the guitar is properly set up, and has new strings on it — these will sound brighter, sustain better and help the intonation. Poor intonation can be a real problem: It means the guitar will be in-tune when open strings are checked on a tuner, but out of tune when played further up the neck. The only way to deal with a guitar with an intonation problem is to tune it in the position where most of the chords are played: If this doesn't work, you'll have to get the instrument set up by an expert, or get another guitar.

DI Sound

Most modern acoustics have some sort of piezo electric pickup built-in, usually in the bridge. Without wishing to annoy the manufacturers, I must say that I find the DI sound inferior in most circumstances to the sound you can get with a mic. The DI sound comes into its own in live situations (gigs) by greatly reducing feedback problems; so if you're trying to record a whole backing track live, and you haven't got enough space to get separation on the acoustic, then use the pickup.

You may, of course, feel that the slightly plastic electronic sound of the pick-up is appropriate to the track you're working on — if so, then fine — here's what you do.

Set the guitarist up in the your live room (so you're not confused by hearing it acoustically at the same time). Plug the guitar into a DI box and get a reasonable level. I don't normally use much EQ at this stage, a bit of a boost at the top (15-20Khz) for sparkle, perhaps a bit of a cut at about 3-6Khz to remove some of that electronic zing, and a cut of a few dBs at about 0.5Khz to remove boom. You'll just have to use your ears here, don't overdo the EQ (you might not be able to undo it later). Always compare the new sound with the old, to make sure you're actually improving it.

It goes without saying that the next stage is compression. Assuming that the input level is about right, patch a compressor across the DI channel, set the threshold so all but the softest signals are above it, set the ratio so the gain change is about 3-5 dB, and adjust the make-up or output gain to get the right recording level. This will give you a light, bright, punchy sound that comes through okay, but doesn't take up too much room in the track.

Mic sound

Recording acoustics with mics can be quite a complicated process, involving esoteric methods such as multiple mic set-ups and valve compressors and equalisers costing thousands of pounds. Without wanting to dismiss the merits of taking time and care over recording, we should merely observe that some of these arcane practises could appear to do as much to enhance the mystique of the recording engineer as to achieve the perfect sound. Suffice it to say that you'll be able to get a great sound yourself using one reasonable mic and a compressor. Here's how.

In my introductory article on recording guitars I spoke at some length on microphone types and applications; I don't want to bore you by repeating myself, but if you didn't see that piece, you'll need to know what to use. If you've got a large diaphragm condenser (Neumann U47, U67, U87 or AKG C414, for example), then this is the favourite. The next best is a small condenser (Neumann KM84 or AKG C451); if you don't have any of these, then use your favourite vocal mic: An SM58 or similar will give perfectly acceptable results.

Get the musician set up, seated on a comfortable, armless, squeak-free chair (this helps keep a consistent distance from the mic, and thus a consistent sound level). Put your microphone about 18 inches from the guitar, pointing halfway between the soundhole and the octave fret, and ask the guitarist to play the part. I always add about 3-6dB at about 15KHz for silvery sparkle, and usually the same at about 5KHz for brightness. You don't want to overdo the top end, or the sound may become thin and scratchy. Then boost the lower mid and sweep the frequencies, listening for that frequency that makes the guitar sound muddy and boomy — it's usually about 500-600 Hz; when you find it, you should cut it by about 6 dBs. Finally, try rolling off a little of the bottom end; if you've got a high pass filter, set it at about 80Hz.

All these figures are approximate, and depend on the instrument, the acoustic environment, and the microphone type and position — you must use your ears, experience and personal taste. Frequently check the EQ'd sound with the flat sound, to see if you are actually improving it (but remember that louder usually sounds better). Don't be afraid to take the time at this stage to try different EQ settings — it will pay off. When you've got a sound you're happy with, set the appropriate recording level and put a compressor across the signal; set it in the same way as described above for DI sound.

Because an acoustic guitar is intrinsically fairly quiet, and you're using a very sensitive mic with a lot of top-end boost and compression, extraneous noise can be a real problem. I don't find gates much use, as the unwanted noise usually occurs at the same time as the music. Loose string ends, rattling coins and jewellery, audible foot-tapping and clicky plectra can spoil an otherwise good take: you'll just have to listen out for them, and deal with them physically; sometimes a harder plectrum will sound less intrusive, snip off string ends, remove jewellery and empty pockets.

Another similar problem that can be very awkward is what is known as can-spill. If the guitarist has anything in their headphones that won't be part of the final mix (prime example: a click track), some of this spills out of the cans and gets picked up by the mike. If there is a passage in the song where the acoustic guitar stops and the sound decays naturally, an intensely irritating click becomes apparent. Sadly I can't offer any miracle cure for this; (readers' suggestions welcome) always use closed type headphones, keep headphone levels as low as practicable, and be prepared to ride guide track levels during recording.

Multiple mic set-ups

The previous approach will give you an excellent acoustic sound for a track where the acoustic guitar is only a part of the instrumentation; it will do its job and come through the track without taking up too much space. If, however, the acoustic is going to be the main or sole instrument then you may need a bigger, fuller sound. One way to achieve this is to use two or more mics.

Set the guitarist up as above; place the the first mic (ideally a large condenser) about 2-3 feet from the guitar, pointing at the sound hole. This needs to be EQ'd to give a slightly fuller but less bright sound than we got using the method above, so don't add as much top, and don't remove as much low mid and bottom. Use only gentle compression (3dBs maximum): We're going for a fuller, more natural sound from this mic. Then put a second mic (preferably a pencil condenser) about six inches from the neck, pointing at or near the 12th fret. This should be EQ'd more extremely than the single mic — give it lots of high top, and pull out most of the low mid and bottom end: you can afford to compress this fairly brutally — use up to 10dBs of compression.

The next stage is to combine the signals from the two mics. Route them both at about the same level to the same group, and listen to the combined sound. Then try reversing the phase of one of the mics (when recording with more than one mic, there is a rule that says if the distance between the two is less than three times the distance of the nearer one from the source, then phase cancellation can occur). The combined sound will be appreciably richer and more natural when the correct phase option is selected: Once again, you'll have to trust your ears on this one. Finally, adjust the balance between the two signals to taste — usually more of the warmer signal from the more distant mic, and less of the bright thin sound from the close mic.

Stereo guitar

The method described above will give a warm, full, natural-sounding track that has plenty of hi-fi detail and precision. It will, however, be mono. If you need a stereo sound, there are two different approaches. If it's mainly a rhythmic strummed part, then my preference is to record the part twice (doubling), and panning the resultant two tracks fairly hard left and right. The slight discrepancies in rhythmic emphases will sound good. You can try using a very slightly different varispeed setting on each take: This will give a great natural chorus effect, but beware of tuning problems.

For an even fuller sound, get the guitarist to double the part as normal, then, using a capo, play the part in a higher inversion. Say the song is in the key of G, put the capo on the fifth fret and play the song in D, and double this as well. The resulting four tracks will sound brilliant together. Another, more involved, method is using so-called Nashville High Tuning: this requires another guitar strung like a 12-string but without the main 6 strings, so effectively you're doubling the lower three strings an octave higher, and the higher three strings in the same octave. You get a 12-string sound without the grief of having to play it.

It may be, however, that doubling is not appropriate. Say the guitar part needs to sound like a single performance, or it is less rhythmic and involves picking, yet you still want a stereo sound. So, keeping your close mic set-up, put two good mics (ideally large condensers, but you could try PZMs) at least six feet apart, and at least six feet from the guitar, and pointing towards it. The right EQ for these will depend largely on the acoustic qualities of the room you're using — the sound should be bright without being thin, and full without being boomy.

Now, set up a stereo group on your desk, route the close mic signal into the middle, and put the signals from the other two mics hard left and right. Starting with just the room mics, gradually bring up the close mic till you get enough sparkle and precision in the sound. Finally put a stereo compressor across the stereo group, and adjust to taste — you probably don't want too much — 5dBs should be plenty.

Sorry to be so long-winded this month. Some of this may sound a bit complicated, but do persevere — getting good sounding results from natural sources is both highly challenging and highly rewarding. By following the methods I've outlined above, you will, with slight modifications, be able to get good results. Remember, acoustic recording is most creative, just use your ears, enjoy yourself and have fun.

Listen with Brian on this month's CD, as he takes you through the theory and practise of acoustic guitar recording.

Errata: This is actually track 11 on the CD, not track 4.