Mic Placement When Recording Impulse Responses

by David Gauger II

"Shofarsogood"

It goes without saying that there are two philosophical approaches to recording:

1. Classical - the recording you make simply documents what is happening in the concert venue. Little attempt is made to change the sound of what you're capturing. The recording is not so much of the artist or orchestra, but of how the room treats the sound of the artist or orchestra. In other words, you're attempting to record the room that the concert exists in. This follows the Tonmeister school of thought.

2. Pop - The idea is to minimize or even eliminate the room sound and just record the artist or instrument. Ambience is added later, usually electronically with a reverb box (or convolution!) of whatever type the flavor de jour is. True stereo is often difficult to achieve since many things are recorded monophonically (one mic per instrument or sound).

To capture an impulse response (IR) of a room, think classical recording. After all, what we want is the sound of the room. Quite a few books have been written on the subject of classical recording techniques and I'll list some books at the end that you may find helpful. In the meantime, use true stereo (and surround) mic techniques in order to capture a room in stereo or surround as the case may be. To make this article manageable, let's stick with stereo mic techniques.

To cut to the chase, when impulsing a room, just set up your mics and make all technical choices much like there is an orchestra on stage and you're doing a classical recording. Recording IR's then is mainly about mic choices, distances and stereo arrays (geometries). Surround IR's are similar, only the issues are about surround arrays of mics instead of just pairs of mics in stereo. Lot's has been written about stereo micing and a quick google will turn up all sorts of sources for this type of information.

God gave us two ears and the ability to hear in stereo and even surround. Researchers have identified 3 different mechanisms our ear/brain system uses to localize sound. We need to know about these to take advantage of them with our mic geometry choices.

If a sound comes from 45 degrees to our right, the sound hits the right ear earlier than the left. This is called the ITD (interaural time difference). It is also slightly louder in our right ear. This is the ILD (interaural level difference) sometimes called the IID (Interaural Intensity Difference). The sound also bounces off the folds of the outside of our ears (the pinnae) causing comb filtering that colors the sound. Our brains decode the coloring to detect angle, height, and to differentiate between the front and back locations of the same angle. Researchers have measured the comb filtering and the resulting measurements are called HRTF's (Head Related Transfer Functions.) Throw that phrase around at your next party! The ear/brain system uses all three localization methods at different points in the frequency spectrum. Some are more effective at certain frequency areas than others.

You must use two of the same kind of mic.The more closely matched your mics are the better the stereo aspect of your recording will be.You cannot get around using two of at least the same model of mic. It will not work to use a Shure SM57 (dynamic cardioid) and a Rode NT-1 (condenser cardioid) as a stereo pair. Won't work. Give it up. On the other hand, many manufacturers will sell you a pair of like-model mics they have matched using various methods of their own (graphs of frequency response, listening tests, etc.) for a fee. I've never wanted to pay the fee for this service, so I just use two of the same model mic off the shelf and live with the fact that they may not be exactly matched. So far, the consistency of even the inexpensive mics I use allows for this, at least to my own satisfaction.

In a nutshell, the mic geometries that capture only amplitude level differences (ILD) only are called "coincident" mic pairs indicating that the capsules are as close together as physically possible. Examples are Blumlein, XY, and Middle-Side types. There are many great sites on the web that describe these stereo mic configurations, and I'll not duplicate their efforts.

For reverbs, it's best to focus on geometries that capture the ITD - time difference between the ears as these tend to sound much more spacious, a good quality for reverbs to have. After all, if the IR is overly spacious, you can always pan the reverb returns more to the center. Geometries that capture ITD include ORTF and NOS (near coincident), Decca Tree, OCT, OSS, spaced omni's and spaced cardioids. In many cases, I personally prefer rear-facing spaced cardioids because they give the wide feeling of a spaced pair but capture mainly reflected sound from the back wall or corners. The reverb sounds deep and full. To my ears the rear-facing spaced cardioids give more of the room's personality to the sound than the other pairs, though for some people it might be too much coloration.

When using a spaced pair like omni mics in a classical setting, conventional wisdom says the distance between the mics depends on the physical width of the musicians you're recording. Typically, this would be at the 1/3 and 2/3 width points. For example, if a small orchestra were 30 feet wide, your spaced omni's would go 10 feet in from each side putting them at the 1/3 and 2/3 width of the orchestra (i.e. 10 and 20 feet from the left side of the orchestra). I read somewhere that when you're trying to capture ambience, it's more effective to move the mic's even farther apart than 1/3 and 2/3. If you look at the recording notes from some of the IR's on this website, you'll see wide distances between the spaced mics for this reason. Maximum spaciousness is the goal.

Almost all of the impulsing locations on this site have had a stage. As a result, the actual impulse sources have been located on the stage as opposed to the audience or back of the hall. The idea of a forward (front) dominated listening environment is still typical for most recording projects. All of these IR's have been recorded with mic distances, forward orientation, and geometries chosen to maximize this aspect of the sound.

In general, when your mics are close to the impulse (gunshot at the front of the hall), level-wise you get a lot of the impulse spike and little (by comparison) of the reverb tail. In the back of the hall, the reverse is true: not only is everything softer, but also the initial spike is closer in level to the reverb. The ratio of spike to reverb changes. I find that the most pleasing IR recordings are those where the mics are placed no closer than 1/3 of the way back in a hall. This is farther back in the hall than most classical recording situations where the mics might be 10 feet or less back from the closest musicians, although the mics might be flown high in the air. I also typically place my farthest pair about 3/4 of the way back in the hall or maybe farther. If your mics are really close to the back wall, rear-facing might not be your best choice but you might get away with front-facing cardioids - follow your ears.

It seems that every IR, no matter what the distance, has some musically useful qualities to it. Each distance has a different sound - a shift between presence and ambience, coloration and envelopment. Each mic geometry sounds different, too. With so many choices nearly all of them sounding good, which IR is the "right" one? My experience is that they're all right in some sense or another. (How PC can you get?) So my solution is to record as many geometries and distances as I have available equipment. Lately I've recorded 5 or 6 stereo pairs to let the person using the IR's make the choice. Maybe this is too much and buries the user in a blizzard of choice, but in the end, I think more choice is better than less.

Most mic stands will typically will go 8 or 9 feet high so that's the height at which most of the IR's on this site have been recorded.

Using small diaphragms for ORTF and NOS geometries is a good idea so off-axis coloration won't be a problem. Large diaphragm mics are notorious for having off-axis coloration but this is not a problem when they're used in spaced cardioid configuration. In fact, many inexpensive large diaphragm condensers are noted (negatively) for having a pronounced presence rise, but this can be considered to be an advantage when using them for back of the hall rear facing spaced cardioids. The presence rise helps the presence of the reverb as well. Yup, cheap Chinese knockoffs and East European mics do a respectable job recording IR’s. Leave your Neumann U87ai at home.

You can leave the Avalon mic pre's at home, too, along with the Urei 1176. Many of these boxes add pleasant and desireable color to the sound of a vocal, but we're not recording vocals. We're just trying to document what is there in the room, not change it, and with reasonable care setting record levels, limiting isn't necessary, either.

Quiet mics with reasonably low self-noise (less than about 16 or 18 dB) are good for impulsing as signal to noise ratio is a problem since the reverb decays down to nothing. The reality is that most rooms are not very quiet. Between rumble from cars and noise from air handling equipment, many of the bits in your 96kHz/24 bit recording will go to faithfully reproduce…the ambient noise in the room! Some of the toughest problems are to get the all air handling equipment turned off. Many times you must contend with computer controlled air handling where only one or 2 people can access the computer to override normal settings to get the room quiet enough. Sometimes the blowers will be off in the room you’re impulsing but you’ll still get noise from blowers in the next building!

The great majority of the time, the noise level in a real room will be much higher than the noise floor of even modest equipment. Useable signal-to-noise ratio is difficult to achieve in real-life situations. That is why since the ambient noise is so high, we must resort to extremely loud sounds (e.g. gunshots) to get a reasonable signal to noise ratio. While you might be recording at 24 bits, 8 or 10 of them record levels below the ambient level in the room, which makes your 24 bit recording effectively only 14 to 16 bits of useable signal level. Typical ambient levels in the spaces I’ve implused are at about –60 to-65 dB, if I recall correctly. Since 16 bits has a theoretical max. dynamic range of 96 dB, we’re OK, even with 16 bit recording!

After the equipment has been carried in, clap your hands a few times as you walk around to get a feel for the room. As a general rule, the closer the mic pair is to the stage, the more you should tend to use near coincident mic geometries to get locational accuracy. The farther away from the stage, the more the goal is spaciousness and envelopment so go more for spaced omni’s and spaced cardioids. The more diffuse the room sounds during your clapping, the closer you should put the nearest pair to the stage. After that, just space the mic pairs out 8 or 10 feet back to the next pair, always left/right centered to the stage. When all is said and done, it's good for the mics to be spread all around the room both in distance from the stage and also distance between spaced mics. That way you have more complete “blanket”-like audio coverage of the room.

Try putting, both an ORTF pair and spaced omni’s in a line the same distance from the stage. That way you can have either accuracy or spaciousness, or both, at that distance. Once the mic locations are chosen and the mic lines run, measure every distance you can think of because you'll want to know what you're listening to. Of particular interest are the dimensions from the stage to each pair and the distance between the spaced mics.

Here's a tip: set the digital recorder up in the center of the microphone “field”. That way you can get away with fewer mic cables since none of the mics is very far away from the recorder. After all, there's no audience to worry about.

Make your mic selections, geometries, and placements as if you were doing a classical recording. After all, we're after the sound of the room! Use relatively quiet mics, especially those towards the back of the room that receive less energy from the impulse than those closer to the impulse source. Emphasize near coincident and spaced geometries over coincident mic pairs because most people like the spacious quality these geometries provide in the resulting IR's. Have fun!

2. The Microphone Book by John Eargle