The Art of Noise: Game Studio Recording and Foley
By Robert Stevenson
Gamasutra
September 22, 2000
URL: http://www.gamasutra.com/features/20000922/stevenson_01.htm

Undoubtedly you've encountered a situation similar to this: For months you've been eagerly awaiting the next great game from Design Team X, drooling over the ads featuring near-photorealistic screenshots, rereading dozens of times all the flattering previews found in the magazines and web sites, and even more carefully scrutinizing the publisher's glowing copy on how this new game will be the greatest thing ever to grace your computer. Finally the day comes. After a hasty call to the store, you tear off at lunch to plop down your hard-earned money for this holy grail of gaming greatness. You install and begin to play. The graphics are truly phenomenal and the gameplay is starting to look exciting. Vile aliens and brutish dragons are off in the distance descending rapidly on your starting point. Deftly you leap to the closest pillar, grabbing the Mega-LFG and preparing for the initial onslaught. The enemy draws nearer. You raise your weapon. They close in almost revealing their beady little eyes. You wait for the perfect shot pressing the trigger on your trusty mouse, ready to unleash hell incarnate on the infernal foe, when out from your Ultimate Psychoacoustic Dolby X Satellite Speaker Array struggles a puny little "pop." Ugh…The Mega-LFG sounds like little more than a chintzy peashooter.

No matter how heated the ensuing battle in the back of your mind, the game is ruined. It feels fake, not on par with what you expected. Even the weapons seem wimpy, meaning the combat must be unbalanced to give the enemies the edge. Of course that must mean the game design isn't really that solid. And without good game design the latest release from Design Team X lasts only a few days on the hard drive and a lifetime collecting dust on the shelf.

On the increasingly level playing fields of 3D acceleration, elaborate input devices, and broadband capability, game companies are working harder than ever to find ways to differentiate themselves from one another. While design and art play a major role in this process, sound is also an important factor in a game's success or failure. In fact, it's increasingly rare to find top-selling titles that don't also feature top-notch audio design. So if you're having trouble making your game's sound environment bigger and more distinguishable from your competition's, this article is designed to help.

Sound designers have myriad tools and techniques at their disposal to create a complete audio environment for their game. Sound effects libraries are often the easiest starting point for a design job and are available in dozens of flavors, from big production houses down to small, one-man shops. In general, stock sounds are available on individual CDs or in complete sets with a common theme. Some of the more progressive sound providers offer download capability from their web sites, allowing you to grab samples even while pulling an all-nighter.

In general, prices for sound CDs and libraries match the quality of the material on them. Cheaper libraries will tend to be just straight (sometimes monophonic) recordings of the source material. Noise and background leakage may even be present. Nicer libraries will be clean recordings, occasionally even enhanced a little, and feature good indexing, allowing you to find what you need quickly. The best libraries will have been pieced together by professional sound designers.

Sound generation from an electronic source, such as a synthesizer or a software package, is another route to take. Sound designers working off of traditional white or pink noise functions, which provide an even noise distribution over a frequency range or octave values, have often found it the most flexible way to sculpt a complex sound. Because synths provide a clean source from which to work, recording issues are also not a concern. Often, intermixing or triggering a completely artificial sound with studio or field recorded samples can produce convincing results.

Field recording is somewhat similar to studio recording, except that in many cases you will not be in direct control of the source material, but will rather be working around it. Field recording can be especially useful for recording large scenes and ambient waveforms. In many cases just having a portable DAT and microphone will do, but sometimes a more elaborate setup is needed. Other issues to consider when field recording are sometimes less obvious. For example, at a military base, the recording of some equipment may not be allowed without permission. Other places may have restrictions regarding free movement or the kinds of recordings you can make. Even worse, some environments may have electrical equipment that will interfere with your gear's recording capability.

While not a replacement for other sound design techniques, in-studio recording of sound effects (called Foley) saves both time and money spent searching for the right material elsewhere. For cutscenes or streamed audio clips, Foley can be entirely comprehensive and extremely fast. Additionally, material can be created that is perfectly matched to your production, not shoehorned in from an audio CD. Even today, it is not uncommon to hear some of the same tired sound effects reused in game after game and in commercial after commercial. The determined walk-step of your gun-toting hero can be cast a certain way, while the sneaky footsteps of the villain he is about to fight can be cast in another. Most game designers would cringe if they thought the same piece of game art from the competition was being reused in their game. Why should sound be any different? Finally, Foley techniques provide an excellent excuse to junk up the development studio with all kinds of fun, eclectic props (see sidebar, "Foley Props").

This article is intended for sound designers who are looking to go beyond the "sound via CD" techniques commonly overused in game production. The focus is on studio recording within typical game development budgets. Experienced sound designers already working in upscale studios may also find this article a good refresher.

Setup

Although in-studio recording is a fairly involved process, setting up correctly before a session can be timesaving. In fact, a lot of things ought to be done before the first audio take occurs.

There are no hard and fast rules for budgeting sound design, other than to make sure you do actually budget for it. Too often on small projects it's listed as something the sound programmer or the music guy can do on the side, or in some cases it's forgotten altogether. It is also good to tackle sound development early enough in the project to make sure it nears completion long before the headaches of trying to master happen. Using an outside contractor for your audio development is recommended if you can't do the work appropriately in-house. Bringing them on the project early will also yield fewer hassles and better results, as there will be more time to make everything perfect.

The specific time requirements for sound effects development will vary from designer to designer and project to project. Variables such as how complex the sound requirements are, the sophistication of the audio equipment being used, and how much pre-existing library material is already on hand also enter the equation. For a typical game development situation, a good rule of thumb is to budget about half an hour per incidental effect (minor or background-type material) and a full hour for major elements (foreground and repeated sounds). Foley work can expedite this process, particularly for minor effects.

Budgeting equipment is another issue to consider. Starting from scratch, you can expect to put together a simple game-capable recording studio for both vocals and effects for under $5,000. Adding high-end, feature-film-quality equipment or a multi-channel digital recording system can run you quite a bit more. If your sound budget is tight, consider renting sound equipment rather than purchasing it. Items such as microphones, mixers, and DATs typically rent for nominal rates and will often be cheaper on weekends.

When planning an in-house recording session, creating a comprehensive sound list or an audio storyboard for the game is a good starting point. Generally, these kinds of lists are based on unit tables or level descriptions and should be broken out by sound type. Keep in mind, sound lists are just as useful for finding out what you don't need to create as they are for what you do. From the sound list, it isn't difficult to break sound groups into rough production categories of how each waveform will be built. Foleys are generally best for "personal" sounds attached to characters and creatures, or specialty effects that require some individuality to set them apart in the game. Cutscenes or animations are also prime candidates for custom recording work.

Outside of the sound list, you will need some type of recording report to track what actually happens during the recording session (see Figure 1). While session notes can be scribbled on a piece of scrap paper, the time it takes to create an official comprehensive reporting system will be more than made up when you are ready to edit and mix down your effects. A typical report will have the sound or sound group name, along with take numbers, the presence of the right and left audio channels, and some basic equipment setup information. The recording report should also include some type of indexing scheme that matches your equipment or saving process. DAT index numbers, minidisc tracks, file names or SMPTE Time Code will work for the task. Also be sure to leave room for miscellaneous notes on each take. You may know right away when something is particularly good or bad and can make note of it. With a well-planned recording report and a comprehensive sound list available you just need something to record.

Figure 1: A recording report tracks what happens in the studio, and is an excellent reference later on during editing and mixing.

Through the years, sound designers have relied on all manner of noisemaking objects (dubbed "props") and tricks to create the sound they are looking for. When the actual object needed for recording isn't available or is too unwieldy for the studio, creative improvisation often works best.

It's a good idea to start a prop box filled with all kinds of noisemakers long before a recording session takes place, preferably at the start of a project. Props can be almost anything from baking soda to stalks of wheat; the only limit is your imagination. In situations where a prop might be destroyed during the take, such as the shattering of a glass, make sure to get multiples for practice or mistakes. It's also a good idea to have on hand various wearable prop items in your collection of goodies. The sound of a trench-coated hit man drawing a pistol from his pocket is going to be a lot more convincing if a trench coat is actually worn in the take. The same notion applies for shoes, armor, helmets, jewelry, and specific types of fabric, such as silk and wool.

Foley pits are another addition to your recording area if you anticipate needing to do footsteps. Essentially these pits are lowered sections of floor filled with different walking surfaces. Sand, rock, water, and tile are all common pit fills. A simple inexpensive pit can be constructed out of a deep two-by-four frame (see Figure 2). Materials you might work with, such as concrete, flagstones, metal grating, and even sod, can be purchased at almost any major hardware store.

Figure 2: An example of a "budget" Foley pit setup, ideal for recording footsteps on a variety of surfaces.

The space in which you record a sound is almost as important as the sound itself. If you don't have access to a sound recording facility or are working within a tight budget, almost any moderately sized room, preferably one that is isolated, can be adapted to the task. The most pressing concern with a recording room is controlling the acoustics and, if possible, using them to your advantage. Problems generally arise from sound bounces, where a sound wave traveling from its source is reflected off a hard surface, such as a table or wall. In a small space, these reflections are usually not perceptible to the human ear; in larger spaces they can manifest themselves as an echo or multiple dispersed echoes, called reverberation.

These alterations to the sound, sometimes referred to as coloration, can often be dramatic, not only betraying the characteristics of the recording space, but also modifying the fundamental characteristics of the sound itself. Luckily, a little knowledge of sound waves and the physics involved can help eliminate such problems before they start.

Sound waves are normally created through the elastic compression (the rapid pushing together) and rarefaction (the rapid pulling apart) of air molecules (see Figure 3). While one phase of compression and rarefaction is called a cycle, the physical distance each cycle covers (noting that sound travels at roughly 1,130 feet per second) is referred to as its wavelength. The number of cycles that occur in a single second are measured as a frequency in hertz (Hz). Humans with exemplary hearing can detect frequencies from about 20Hz to 20kHz, although you'd be hard pressed to find any modern audio standard delivering the entire range.

Figure 3: How sound waves work. A little knowledge of the mechanics involved will help make your recording sessions successful.

Because sounds are physical waveforms moving through air, they are free to intermix. When two sound waves encounter one another, their interference can be termed either "in phase," where both waveforms are perfectly aligned, or "out of phase," where the two waveforms are shifted (phase-lagged) apart. The result is a new waveform based on the sum of the displacement of each contributing waveform (see Figure 4). In the two most extreme cases, perfectly in phase and exactly 180 degrees out of phase, a pair of sound waves will double in amplitude or completely nullify each other, resulting in no sound. Either case is a recording mess.

Figure 4: Different waveforms can combine with varying results. On the left, two different waveforms combine to produce an "in phase" resultant, on the right, the resultant is "out of phase."

Sound waves have other important properties to consider when setting up a recording session. Each waveform's compressions and rarefactions also have a strength known as the intensity. The spread of values from minimum to maximum intensity is the sound's dynamic range. When a sound wave radiates from its source, it starts with a uniform intensity pushing on a volume of air. As it continues to travel away from its source, the expansion continues, but it has to push increasingly large volumes of air to achieve the same result. Without amplification, the sound's acoustic pressure dissipates, decreasing its intensity. The same behavior can be observed in any waveform outside of a vacuum. For example, as the shock waves from an earthquake spread from its epicenter, the area affected increases while the relative strength of the shock waves declines.

Sound pressure varies inversely with the square of the distance from its source. In numeric terms, an audio wave's intensity drops by six decibels (a logarithmic measurement of a sound's loudness) with each doubling of its radial distance from the source. This exponential relationship is known as the inverse square law, and keeping it in mind during any recording process can be extremely helpful. By increasing the distance from your microphone to the closest source of reflections (at least two to three times greater), you can dramatically diminish the intensity of any audio coloration (see Figure 5).

Figure 5: The inverse square law can help you diminish audio coloration through proper microphone placement.

Of course, providing increased distance from the source of reflections is not the only thing you can do to ensure a clean (sometimes called a "dry") recording. Soundproofing the recording space from outside interference is a good start. While full sound studio setups can run into the millions of dollars, small changes to any space used for recording can make a big difference. Unless they were designed for acoustics, rooms with windows are not advisable. Any doors in the recording space should be replaced with the solid-core variety and the walls should be soundproofed with insulation. Felt-padding or weather-stripping doorframes and air vents can also help eliminate vibrations caused by air pressure variations in a building.

Music supply companies sell wide varieties of diffusers, used for dispersing an audio reflection evenly, and dampening tiles, used for trapping sound. Hanging these on the walls of your recording space near the closest reflection areas to your audio source or in the direction of the microphone's pickup (see below) can work dramatically well. If you're on a very tight budget, consider purchasing acoustic blankets. More often found at theatrical production houses than music stores, they can be hung or draped about for adequate effect.

Microphones

Perhaps the most important piece of hardware in the recording room is the microphone. While there are literally hundreds of manufacturers, types, and styles, they can be broken down into a few basic categories. A good Foley setup will typically have a wide range of microphones available. Sometimes it may even be useful to presort microphones into smaller groups that complement one another.

The process of taking compressed waveforms of air molecules and turning them into electrical signals is the function of a microphone's pickup. Pickup mechanisms are often located near the tip of the microphone and are usually encased behind some type of acoustically transparent protection.

Dynamic pickups are older, simple electromagnetic devices that operate off the vibration of a polyester diaphragm. As pressure waves strike the diaphragm, an attached conducting coil moves back and forth axially along a stationary magnetic field. The movement of the coil through the field induces an electric current directly proportional to the amount of vibration in the diaphragm.

Dynamic microphones make great general-purpose devices, good for recording low frequency or especially loud material. Good dynamic mikes will typically have a frequency response range from 50Hz to 17kHz. Musicians often use them to record drums.

Condenser microphones, sometimes called capacitor or electret mikes, are a more recent invention and operate differently from their dynamic cousins. The diaphragm for a condenser microphone is some type of charged metallic surface mounted adjacent to a fixed, oppositely charged plate. As sound waves vibrate the diaphragm, the gap between the plates alters, allowing some electrons to jump from one plate to the other, which varies the capacitance and charge, yielding a small electric current. This current is so weak that it must be amplified before it exits the microphone. Therefore, condensers have a small preamplifier mounted in the mike body just behind the pickup. Because of the diaphragm mass and sensitivity differences, condensers absorb a broader frequency spectrum than dynamic mikes, ranging from as low as 20Hz to as high as 24kHz.

With traditional condensers, there is a need for a separate (sometimes called "phantom") power supply to actively charge the diaphragm plates, which makes them a little cumbersome for fieldwork. Electret condensers, a special subtype manufactured with a chemically polarized diaphragm, do not require additional power and so avoid this problem. Condensers of all types are not the most rugged microphones, and care should be taken when handling them. Nonetheless, condensers make excellent Foley mikes, allowing the widest range of sound to be captured, giving your work an accurate and rich feel.

In addition to the pickup, microphones also have a sensitivity pattern that governs the area from which they can gather aural input. As their name suggests, omnidirectional microphones have a spherical field pattern radiating out from the diaphragm (see Figure 6). This makes omni mikes useful when background noise or acoustics are intended to be in the mix.

Figure 6: An omnidirectional microphone has a spherical field pattern, ideal for when background noise is meant to be part of the recording.

A bidirectional microphone picks up source material from both the front and back equally well (see Figure 7). Sound coming from the sides will be only minimally captured, if at all. This configuration is sometimes dubbed "figure eight" because of the shape of its field pattern.

Figure 7: A bidirectional microphone picks up sound equally from both the front and back of the mike.

Unidirectional, or cardioid, microphones are sensitive front-diaphragm devices usually favoring higher frequencies in their directivity. They are useful in stage and voice work, where they can be aimed at the sound source, and pointed away from the audience (see Figure 8). In Foley work, cardioid microphones are very common because they offer directionality with the freedom to move around.

Figure 8: Cardioid, or unidirectional, microphones are used commonly in Foley work.

Hypercardioid microphones feature longer mounting tubes and increased directionality over traditional cardioids. Taken to the extreme, these microphones are called supercardioid or shotgun mikes. These rifle-like microphones have a pickup pattern that matches their name - essentially straight out in front (see Figure 9). Because of their high sensitivity, they are excellent at picking up sound within a very narrow field while eliminating nearly everything to the sides and rear. This can be incredibly useful in situations where the sound source is statically located in front of the mike, such as using guns or hand prop manipulations. However, in a situation with a moving source, such as an actor or a large Foley prop, shotgun mikes become unwieldy and difficult to track with, even for veteran operators. It's also worth noting that no matter what you may have seen on TV, typical shotgun microphones don't have any more range than their cardioid brethren.

Figure 9: Hypercardioids offer increased directionality over regular cardioids, but can be tricky to work with.

From booms to woolies, there are all kinds of microphone accessories available on the market. While most of these accessories are not particularly useful for a typical Foley shoot, a good set of stands and a shock mount are often handy to eliminate vibrations or the occasional microphone trauma. All microphones generally come with some type of mounting hardware but inexpensive plastic clips will do little good in situations where the mike is in close physical contact with a source that can jostle them. It is worthy to note that experimentation with microphone coverings, such as foam, socks, or even condoms, can sometimes produce interesting dampening effects, if desired.

Another important thing to note is that microphone pricing varies wildly, ranging from low-end models around $50 to high-end ones up into the thousands of dollars. While you generally get what you pay for, sometimes the best sound for a particular session can be had with an inexpensive, off-the-shelf mike. It's best to experiment and find a set of microphones that work well with your particular recording environment. Every recording task has a specific set of requirements, and by having a wide selection of microphones from which to choose, you can select one that is best for the job.

Recording

Capturing a sound with microphones is an art within itself. The difference between a "big" sound and something lesser is often due to the placement of the mikes during a take. While dedicated practice will ultimately prove to be the best teacher, understanding some basic principles and common techniques before pressing the record button can help you get off to a good start.

In most cases, for game audio you will want to record monophonically. However, depending on the output specification, you may want a stereo recording. While stereo microphones do exist, it's relatively easy and more flexible to achieve excellent, coloration-free results from two monaural microphones. Knowledge of a few basic arrangement patterns and a little experimentation is enough to get started.

The easiest way to make a stereo capture is to arrange the microphones equidistant from the signal source. This can produce a decent, wide stereo effect on large sources or be useful for capturing multiple points of view, but can cause problems if wave cancellation issues arise, particularly in a dynamic shot. A particular disadvantage to microphones spaced apart is that they will not accurately capture the signal the way human ears naturally hear it because the interaural difference (the difference in physical spacing between channels in a stereo field) is too large. For that, you need to place the microphones in a pickup pattern that makes them sample both the right and left versions of a single, coincident point in space.

Traditional coincident patterns will vary from setup to setup, depending on the microphones used. Typically, two cardioid or (if more ambience is desired) bidirectional mikes will be oriented perpendicular to each other, split at 45 degrees and facing the sound source.

The XY pickup pattern is a common example of the coincident technique using two identical monophonic microphones. The idea is to mimic the behavior of a single stereo microphone by crisscrossing the two independent mikes slightly more than 90 degrees near the tip (see Figure 10), trying to keep them close together. Keeping the mikes arranged in this fashion also helps minimize any phasing artifacts caused by the sound reaching the diaphragm of one microphone before the other. In this case, the left microphone is covering the right side of the sound source and the right microphone the left side of the sound source.

Figure 10: An XY pickup pattern mimics the behavior of a single stereo microphone by crisscrossing two monophonic mikes approximately 90 degrees near the tips.

The ORTF pickup pattern is an extended modification to the XY technique. Simply stated, this pattern is a precise 110-degree separation angle between two microphones spaced 7 inches apart, similar to a human head. The MS (mid-side) pattern is another common arrangement that uses a forward-facing cardioid mike to capture the source directly, and a perpendicularly facing bidirectional microphone to capture any ambiences and reflections. Playing with the input level on the microphones in the MS pattern will produce a narrower or wider stereo field.

In addition to microphone orientation, you will need to select an appropriate distance from the source to record. This is sometimes called the recording perspective. Close-in approaches generally yield a tight focus with little ambience, while distant techniques will pick up the surrounding acoustics. In cases where you are trying to capture a detailed effect, such as a door latch or machine gun shells hitting pavement, it's generally preferable to work close in, keeping the mike gain at a minimum. Equalizing in postproduction on a tight recording can make the sound dramatically bigger. Choosing a more distant positioning may be required, depending on the type of microphone you are using and the size of the sound source. Don't forget the room's acoustics and the inverse square law when planning your setup.

Although you may have already eliminated major sources of reflections within your sound space, sound bounces off small surfaces close to the microphone can often change the tonal quality of the recording. Common offenders are nearby tables or improperly positioned stands that may be in the work area. Luckily, testing for these close reflections is easy, and the fixes are even easier.

By placing a small mirror on the suspect surface and eyeballing it from the microphone's point of view, you can quickly tell if the sound is bouncing off the surface from the source. If the sound source is visible in the mirror, the object is likely causing the unwanted reflections. Reorienting the offending fixture to make sure the prop is not visible in the mirror is the easiest solution.

Now that you've got your props and microphones set up, you'll be tempted to start recording like mad. But before you start rolling, make sure to do some practice runs to avoid obvious problems such as clip-outs, where a sound's dynamic range has exceeded the capabilities of the recording hardware and is experiencing a distorted cutoff. If possible, also record a calibrated half-minute lineup tone to help you, or someone else that needs to work with the session material later, to be able to properly configure the equipment.

Finally, it's a good habit to try to record everything during a session. Often the best takes happen during a practice run. And of course they can be edited in during post-production. If enough planning has transpired, working through even a long sound list should be relatively quick. In fact, the enjoyment of banging around and building effects might end sooner than you'd wish. Just keep in mind that a little ingenuity during the session may prove to be the best weapon in your sound design arsenal. (for additional recording tips, see the "Recording Notes" sidebar.)

Adding Zchwing!

Once you've got days of recordings captured and each effect done six different ways, you're bound to find you're still missing a little something. The process of intermixing, resculpting, and adding highlights to sound effects is often called "sweetening," and doing so can often turn a so-so effect into one that really roars. ("Sound Advice from the Pros".)

Approaches to sweetening can vary depending on the nature of the effect. The background audio track of a cinematic cut-scene might not be worth spending more than a few minutes tweaking the sound, particularly if a music track is also in the mix. However, if it's a foreground effect you're working with, one that will be heard over and over, it may take hours or even days to get the sound just right.

As a rule of thumb, it's best to start by having your multi-part sound clips available in a multi-track editing tool, or in the case of a two-channel editing tool, at least in a mix-ready state before starting any enhancement. For example, a shotgun effect being built for a first-person shooter might start with a cocking sound followed by a trigger effect, and then multiple simultaneous blast effects spanning the entire range of the audio spectrum, each within their own distinct tracks. With the correct setup, all parts of the sound will be on the same acoustical page and in time with each other, making further work easier.

When working with sound effects, it's important to understand the tonality of other audibles they are going to appear with, particularly during cutscenes where there should be a great amount of control over the audio presentation. Avoiding range conflicts caused by having too much acoustic information within a limited part of the frequency spectrum is a crucial task for good sound design. Fortunately, equalization can provide more punch to your effects to push them past the music or vocals as appropriate.

All equalizers are variations on the same theme, producing better-sounding audio by increasing and decreasing the intensity of frequency bands within the audio spectrum. Common types of equalizers are graphic, which allow partial or full octave control over bands within the frequency spectrum, and parametric, which allow for controlled boosting or cutting around a specific frequency.

Using an equalizer effectively takes a little practice, but just as with any other effect you should have a clear idea about what you are trying to achieve before starting. Equalizers are very good at quickly boosting or cutting the bass (less than 250Hz), mid-range (500Hz to 2kHz), or high frequencies (4kHz and over) in a sound. Simulating phone lines or reduced radio transmissions can also be done by peaking between 400Hz and 3kHz. Parametric equalizers in particular are useful for eliminating unwanted noises, such as tape hiss or electric hums, at a specific frequency.

Besides equalization, audio compression is perhaps the most common method used to accentuate recordings by providing direct control over maximizing and minimizing the dynamic range of a signal. Simply stated, compression alters a signal by taking the input volume and moving it down by a factor denoted by a ratio setting. A threshold control makes the ratio change specifically at a point within the dynamic range, leaving everything below it unmodified (see Figure 11). The number of threshold changes and how dramatic they are is referred to as a "knee" setting. Hard knees cause abrupt changes in the compression ratio, while soft knees are small changes in compression staged over a range in the signal. Additionally, how aggressive the compressor acts on the signal as it's processed can be set with attack and decay parameters. Also, a compressor can be output-gained to increase the volume across all parts of the signal.

Figure 11: Manipulating the compression threshold at specific points within the dynamic range produces different results.

Compression tools - particularly ones in hardware - can be daunting, but slowly working through the parameters and making small adjustments while listening to the sound will make the settings much clearer. To start with sound effects, try using compression to simply regulate the high end and eliminate any distorting clip-outs. Settings with a high ratio around 3:1, a high threshold around 12 decibels, and a tight but soft knee are a good base. Bringing the knee up can provide a cleaner round off, while bringing the overall ratio down will compress the dynamic range of the whole signal. Variations on compression include limiting (cutting off signals above a certain level) and expanding (increasing a signal's dynamic range by decreasing low-intensity portions below the threshold). Expanding is also a good way to do smoothly transitioned noise reduction.

Pitch shifting is another common way to improve the tonality of a sound or occasionally just to warp it into something entirely different. Subtle shifting can also be used to fix the timing on sound, for example, to synch it with an in-game animation. With a little practice, even a convincing Doppler effect (the sensation that a rapidly moving sound source, such as a car in a racing game, has a higher pitch when closing on a point in space than it does when moving away) can be created.

Modifying a sound up or down and possibly locking its duration is easily handled with pitch-transposition tools designed specifically for the task. Shifting upwards increases the frequency, making the sound higher. Conversely, pitching down provides a more low-end, bass-like effect. Playing with alternating pitch (vibrato) can provide interesting results. You can also make a particular effect seem wider in a stereo mix; shifting one channel a small amount (generally less than 5 percent) will often work. If it is satisfactory in stereo, mix-down and test it in mono to make sure there are no phasing artifacts and that the results are equally improved.

Pitch shifting is not without adverse effects. Because sound samples are either being duplicated or removed during pitch shifting, it's a destructive process, making it hard to backtrack by simply negating the effect. Additionally, while shifting technology has evolved to a level where interpolation artifacts are generally low or antialiasable (minimized through weighted averages), they will creep in, especially as the sound gets increasingly divergent from its original state. To be safe, pay close attention to the quality of your output, and only pitch-shift a few octaves in either direction.

Effects recorded in the studio often don't feel right when played back in the scene for which they are intended. They feel too close or too confined, like they are playing back just inches from your ears. This is where reverb comes in. The irony is that after you've spent time and resources building the perfect sound studio to eliminate reflections, now you actually want to add some. Reverb is also good for blending edits in multi-part sounds, helping them seem to be from the same acoustical page. Adding reverb is usually a fairly painless task, as most software and hardware units will have decent presets, usually named after materials or architectural spaces. Understanding what is actually happening during the process can help you engage the problem more accurately.

Reverb occurs because sound-reflecting surfaces in a space encourage the sound to bounce around, sometimes half a dozen times before finally reaching your ears. Because some sound waves will invariably travel straight to your eardrum while others take time bouncing about, you will experience the primary sound wave followed by additional, sometimes weaker, echoes of the original sound delayed by just milliseconds. The reflections in a naturally reverberant environment, sometimes dubbed a "wet" sound, will vary not only in timing but also in tonality, based on the shape, area, and absorption qualities of the surface materials in the space. Softer materials provide low-end reflections, while stiff materials bounce high frequencies.

Adding reverb to your effect can take some experimentation. Starting from the top by focusing on the length and density of the reflections is sometimes useful. Working with the intensity decay rate of the reflections is also important in defining the feel of a sound's space. Short decays will suggest small spaces, while slightly longer decays will suggest much larger areas. Nonlinear falloff can suggest an unevenly occluded environment. The frequency range of the reflections is also crucial. Adding more high or low end to a sound's reflections will cause the surface materials present in a sound's synthetic environment to feel significantly different. Most (although not all) reverberators are comprehensive signal processors and will have other delay-based effects incorporated into them like flanging or phasing, canceling portions of the sound to make it feel like the source is moving, and chorusing, making the sound appear to be from multiple sources through minute pitch-shifted delays.

In general, following the ordering of the postproduction effects in this article will have your sound ready for any last-minute finalizing and mix-down. While some steps can be skipped, it's a good idea to work at least with the compression and equalization on each waveform. Although it may work otherwise for some productions, it's usually better to apply any pitch-shifting and equalization before compression to give each process the greatest dynamic range within which to work.

While most studio-recorded effects are one-shot affairs, occasionally they will need to loop during playback. There is often little you can do to create a looping track during recording other than trying for clean, editable breaks in a sound's cycle. During editing, cross-fading multiple tracks of the same sound for loop effects that can't simply be cut in or spliced is a good starting point. For sounds that vary in timbre across their duration, another technique is to reverse a copy and then play them back-to-back.

Panning sound effects is usually not required for 3D games, as most sound work will be incorporated into the game monophonically and placed in stereo by the sound engine. However, if you are working with stereo (or better) effects panning should generally be done at the end of your post-processing. Reverb or any other delay effects can be done at almost any point, although the best results are often found applying it near the end.

The last part of the sound design process is mixing down and doing any necessary format conversion. This can be a complex task for cutscene productions, taking days, if not weeks, in front of a nonlinear editor. However, for most in-game sounds, it's just a matter of combining any multi-part sounds into one singular effect. Keep in mind the final output during the mix-down process. If you have been working with stereo material, verify its mono compatibility. Also keep an eye on the level meters and watch for distortion, particularly if you are additively mixing material with a lot of dynamic range. If you need to down-sample your sound for output, maximize your signal-to-noise ratio with gating or compression before doing the conversion.

To get the most out of the sound memory available in your game, trim your mixed effects to remove any extra dead space prior to and after the main body of the sound. While shaving a tenth of a second here and there may not seem like much, it will add up over the hundreds of effects that make up a game production.

If you do find you have some sound memory and channels (voices) to spare in your game engine, layering sounds is a trick Hollywood sound designers have known for years as one of the ways to create a seemingly richer sound environment. Because hearing is omnidirectional (as opposed to vision) and the human brain can only focus on a limited amount of auditory input, usually one spatial grouping at a time, it can be literally overloaded with sound information, forcing less concentration on specific effects making the environment seem richer. Be careful when adding new sounds at the last minute not to create acoustical mush by overdoing it.

Wrap-Up

There is no single correct way to finalize a sound. In a game studio, it's often best to take a break when you think you've got an effect sequence done, then come back and listen to it later. Closing your eyes during playback so you can focus on the quality of the sound is also a good technique. However, testing to make sure it feels right in the game itself is ultimately the best way to make sure it really has been done satisfactorily.
Sound design is more of an art than a science. Recording and editing fresh material for your game project extends that art even further, providing an almost limitless canvas of aural creativity. But by applying knowledge from practiced sound designers with a bit of physics and a lot of imagination, the entire sound environment of your game can be raised to the next level.

For More Information:

WEB SITES

Equipment Emporium
www.equipmentemporium.com/articles.htm

Acoustics & Vibrations Virtual Library
www.ecgcorp.com/velav

BOOKS

Ballou, Glen M., ed. Handbook for Sound Engineers, 2nd ed. Carmel, Ind.: SAMS/Macmillan, 1991.

Eargle, J. The Microphone Handbook. Plainview, N.Y.: Elar Publishing, 1981.

Mott, Robert L. Radio Sound Effects. Jefferson, N.C.: McFarland & Co., 1993.

Pohlmann, Ken C. Principles of Digital Audio, 3rd ed. New York: McGraw-Hill, 1995.

Yewdall, David Lewis. The Practical Art of Motion Picture Sound. Boston: Focal Press, 1999.

Bio:
Robert is the Vice President of Production at iROCK Interactive, a music game development company in North Carolina. When he's not making games, he's busy playing them. He can be reached for questions or comments at robert.stevenson@irock.com.

Copyright © 2003 CMP Media Inc. All rights reserved.