Spaced Stereo Techniques
So far, I've deliberately simplified matters by recording my audio examples with just a single mic in each test position. However, mono piano recordings are pretty thin on the ground these days, so let's look at what kinds of stereo techniques you might try.
One common tactic is to use a spaced stereo pair. As we've already heard from the 'HorizDisp' audio examples, the differing tonalities of two differently positioned mics will give a kind of stereo effect straight away, but the stereo imaging mostly relies on time-of-arrival differences between them. One of the significant advantages of this approach is that you can use omni mics, with the low-end and off-axis benefits these afford. However, there are also some disadvantages with spaced-pair techniques. The first is that the stereo imaging tends to be rather vague, although a lot of classical engineers and listeners find this sound more musically satisfying, so this could, conversely, be seen as a benefit. What is more clearly problematic in certain situations is that if you don't pan the mics hard left/right you will get phase cancellation between the two signals, which can change the tonality of the sound dramatically. Even if you're not planning on changing the pan settings at all yourself, it still pays to audition any spaced mic pair in mono to check that the sound doesn't completely collapse, as some broadcasters still transmit in mono.
How much difference do small changes of mic position really make in practice? To answer this question, six mics were set up very close to each other and their outputs recorded to create the 'TightPattern' audio files — judge for yourself!Because it's tricky to adjust the stereo width of a spaced-pair stereo recording without its tonality suffering, it's important that you try to get the image width you need while recording, by adjusting the distance between the two mics (the further they are apart, the wider the image). A word of caution here, though, as putting the mics too far apart can cause the sound to bunch up towards the edges of the stereo image, producing an effect often called 'a hole in the middle'. I think you'll find that any spacing above about 1m is liable to start running into difficulties, and at the other extreme, Richard King has sometimes placed his mics as close as 45cm apart for piano recording. One way around the hole-in-the-middle problem is to set up a third mic between the main pair, and use this if necessary to fill out the centre of the stereo picture. This can be a good safety net, but a side-effect is that the left and right mics will both cause phase cancellation with the central mic, so it may take a bit longer to get the recorded tonality you're after in the first instance.
There is nothing to stop you using spaced-pair techniques with directional mics as well, although the mics will need to be placed further away from the piano to achieve the same degree of room ambience. Depending on how you angle the mics, you may also find that a hole appears in the middle of the stereo image earlier than it would with omnis, so you should keep a keen ear out for this. And talking of mic angles, even if you use omnis you might still want to experiment with angling them towards the higher strings if you're after the brightest sound, because the high-frequency response of your mics will almost certainly be best on-axis, especially if you're using large-diaphragm models.
The next set of audio examples shows how some of these mic-placement variables affect the sound. I set up an array of seven spaced mics roughly centred on the position of my favourite mic in the 'TightPattern' setup. All the mics were pointing straight ahead, but angled down towards the centre of the piano, and were recorded to the following files:
- SpacedPair1: A pair of Rode NT55 omni mics spaced at 40cm.
- SpacedPair2: A pair of Rode NT55 omni mics spaced at 1m.
- SpacedPair3: A pair of Rode NT55 cardioid mics spaced at 1m.
- CentreMic: A single Shure KSM141 omni mic placed centrally between the other mic pairs.
A quick note about stereo polarity here: for all the ambient techniques in this article I've stuck with the convention of having the higher strings of the piano to the left of my stereo image and the lower strings on the right, which is fairly common practice in the classical domain where you're usually trying to recreate the audience's perspective. However, in pop productions engineers usually prefer a player's perspective (high and low strings the opposite way round), so I've made the audio examples that way for the close-miking discussion later on. If you'd rather hear things the other way round, you'll just have to go and listen to the files in the mirror...
Other Stereo Mic Setups
If you're concerned, as a lot of broadcasters are, about mono compatibility, then spaced-pair stereo techniques don't really cut the mustard, no matter how good they might sound in glorious stereo. In such cases, coincident techniques are the order of the day, all of which place the microphone diaphragms as close together as possible so that phase-cancellation artifacts are negligible when the signals are summed to mono. Because of the mono-compatibility of these techniques, you're also free to narrow the stereo image of the recording at a later date, simply using the pan controls on the two mic channels.
This picture shows the three different spaced-pair stereo rigs lined up together for the 'SpacedPair' audio comparison files: wide omni and cardioid pairs on the outside and a narrower omni pair on the inside. A further central mic was added to demonstrate one way of dealing with the 'hole in the middle' problem when working with spaced-stereo methods.The main family of coincident setups are the crossed pairs, which derive the stereo image from level differences between the mics. The width of the stereo image is proportional to the angle between the mics, usually called the 'mutual angle' — the larger the angle between the mics, the wider your stereo picture will be, although mutual angles beyond about 130 degrees may begin to open up a hole in the middle of the stereo field. Crossed-pair techniques typically give a much clearer and more precise stereo image, but at the expense of what detractors sometimes describe as a rather 'clinical' sound. The fact that such approaches require directional mics, with their potential for off-axis and low-frequency deficiencies, is another reason why some engineers steer clear of them.
There is another coincident option, though, which does allow you to use omni mics: the Mid/Side (M/S) technique. This also gives you the ability to control the stereo width of your recording without moving the mic setup at all, which is great for situations where the best mic position isn't the most accessible. If you're interested in investigating M/S further, or just need a bit more information on stereo mic techniques in general, check out the links to Hugh Robjohns' articles in the 'Help! I'm Buried In Jargon!' box.
Something of a halfway house between the spaced and coincident methods is another family of near-coincident techniques. These use directional mics with fairly small spaces between them in order to capture both level- and time-difference information about the stereo field. On the one hand, you could say that this combines the more precise imagining of the coincident techniques with the more musically involving sound of the spaced techniques. Alternatively, you could also say that this combines restricted mic choice with the potential for phase-cancellation problems! To help you decide which side of the fence you're on, here are some more audio examples. I left one pair of spaced omnis set up from my 'SpacedPair' recordings, and set up coincident and near-coincident pairs between them. All the mics were angled downward towards the centre of the piano, and you can hear how they sound by listening to the following audio files:
- StereoPair: Two Rode NT55 cardioid mics in a coincident crossed pair at a mutual angle of around 110 degrees.
- StereoPair2: Two Rode NT55 cardioid mics in a near coincident pair at a mutual angle of around 110 degrees and with the capsules 17cm apart (the 'ORTF' standard developed and widely used in French broadcast circles).
- StereoPair3: The pair of Rode NT55 omni mics spaced at 1m, as before.
One final thing to mention before we move on to closer miking methods is that some engineers combine the techniques we've been talking about, in order to overcome potential problems with specific approaches. One example of this would be setting up a single low-pass-filtered omni mic alongside a coincident stereo pair for better low-end response. Another common tactic is mixing in a little of the signal from a widely spaced stereo pair with a main closer coincident pair — the omnis give a more involving ambience and good bass response, while the coincident pair fills the hole in the middle of the widely spaced omni image and gives clearer stereo imaging. Once you get a feel for the principles we've been discussing, these combined techniques present no greater fundamental problem than the simple techniques, beyond the practical considerations of setting up, positioning, and phase-checking extra mics during the session.
Where Should I Set Up The Piano?
If you're using ambient recording techniques for classical-style recordings, it's vital that you find the best possible acoustic for your recording sessions, as it will be all over the final recording. Whatever venue you're in, the question of where to set up in not easy to answer. One thing to factor in is that having the piano right back against a wall or in a corner is likely to boost the low frequencies, because of the way reflected sound from the walls interacts with sound heading out into the room — rarely a desirable effect for classical recordings.
A hard wall within a few metres of the piano can help to brighten the sound overall, by reflecting some high frequencies directly back to the recording mics. If there are no walls close by, the sound of the piano has to go a long way to be reflected back, and as high frequencies don't travel through the air as efficiently as low frequencies, the reflected sound will be duller. However, any strong reflection from a nearby surface may cause some phase-cancellation artifacts at the microphones, which may make it trickier to find decent mic positions.
To give an idea of the scale of these effects in practice, I've recorded the same grand piano with the same omni microphone in four different locations within a concert hall to create the following audio example files:
- LocationCentreOfHall: For this recording, the piano was in a position about two-fifths of the way down the rectangular hall, and was firing down the long dimension towards the microphone and the remaining three-fifths of the hall. This was also the piano position where the majority of the other audio examples for this article were recorded.
- LocationAgainstWall: The piano was moved to the end of the hall, firing down the long dimension towards the mic.
- LocationInCorner: The piano was moved into the corner of the hall, firing out towards the centre of the hall and the mic.
- LocationFiringAtWall: The piano was four-fifths of the way down the hall, firing at the wall, with the mic set up between the piano and the wall.