• Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint
Share this Page URL

Chapter 4. Design > Interactive Mixing and Dynamic Range

Interactive Mixing and Dynamic Range

Currently the two major areas of interest in game audio are interactive mixing and dynamic range.

Interactive Mixing

Interactive mixing—that is, mixing game-audio content while the game is playing as part of the final stage of sound engineering—is a relatively new concept, but it's a process audio developers have wanted for a long time. Without interactive mixing, making a change to a sound's volume, for example, required changing code or text outside the game 99 percent of the time. This meant you had to tweak the sound separately and then test it within the game over and over until you got it right, a process that seemed interminable.

Now we have the means for audio folk to change any number of parameters in real time. Microsoft's Xact (Xbox Audio Content Tool) and Sony's Scream (Scriptable Engine for Audio Manipulation) mixing programs do this, and so does Renderware Studio by Criterion, a commercially available game-sound engine for multiple platforms. Eventually mixing and mastering a game will become a job separate enough to warrant hiring someone specialized for the task. This will happen within the next few years. Let's take a look ahead at what the specialized job will entail.

Ideally, the sound designer will load the game and look at a master list to see what sounds are playing. Indicators include volume bars in the form of decibel meters, one for each sound playing, that flash in the control software or level editor (such as Renderware Studio—or, in the case of the Unreal engine, UnrealEd). These days, many systems still superimpose a text list over the game, forcing the designer to scan the list quickly to see the names, some of which pop up for just milliseconds at a time. (Let that be a nudge to the tool developers out there.)

When the sounds need to be mixed so that one effect doesn't overpower others, you can turn down a sound's volume with the twist of a knob or the movement of an onscreen fader. You can also use faders for equalizing, filtering, and adding effects to the sounds. If your programmer has time to code these controls, definitely make use of him. It'll be well worth the integration time.

While you mix audio, remember to base your judgment on a sound that occurs throughout the game on its global use as well as its local use. This means that if a machine gun sounds too loud at a relatively soft level, don't turn it down if it's likely to get drowned at a level with higher volume or more range. If necessary, create a customized sound object for that level and adjust the volume only for that object rather than changing the sound globally. Now that I've mentioned the words global, local, and range, it's a nice opportunity to enter the next section.

Dynamic Range

Dynamic range is becoming increasingly important in audio development because it deals with overall volume considerations. Ever wonder why television commercials are always louder than the program you're watching? It's deliberate—the advertisers want your attention—but that doesn't stop the experience from being annoying. Films avoid this sort of jarring volume change unless they intentionally use it as a plot device. The same applies to game audio. The difference in volume is described technically as dynamic range.

Unfortunately, dynamic range can't be fixed during gameplay without the help of a real-time dynamics process such as a compressor. This is a consideration that sound designers must incorporate from the beginning. A lot of developers use their highest volume in the game as the same volume played when they turn on a console and the console logo is played. Most consoles have a company or console logo introduction with a volume that rivals that of the THX logo in movies—that would be the maximum volume, and it's usually a good starting point. From there, you need to figure out how voice, sound effects, and music will all interact.

In games, a good way to let players set their own preferences is to give them an options screen with a fader control (Figure 4.6) that will act as a volume control for an entire sound set. Such functionality has been available for a long time, but it doesn't help if an individual sound file is already so loud it would break an eardrum.

Figure 4.6. The user-controlled dynamic range in Unreal Tournament 2004 includes an options screen and volume sliders. Players should have to use these as little as possible.

Beyond setting the game's maximum volume, you can get good dynamic range by using several other techniques. The most obvious (and tedious) of these is doing a good solid interactive mixing pass before the game ships. This involves, yep, playing the entire game and correcting any sounds or music that are too loud or too soft, fixing equalization and effects, and so on—polishing whatever audio needs it.

Another method, which can be used in conjunction with the overall audio pass, is setting appropriate ranges in each sound category. Todd Simmons, a sound designer at Ion Storm, suggested this idea while we worked on Deus Ex: Invisible War, and it's brilliant: Set levels for voice at the highest range—from, say, –13 dB to –6 dB RMS. The RMS, or root mean square function, is a way of maximizing the sound level (also known as normalizing) across the entire waveform rather than maximizing the peaks. Don't worry about what RMS means technically; just know that it's an option in programs such as Sound Forge (Figure 4.7). You'll find the RMS option in the Normalize function.

Figure 4.7. Sound Forge 6 allows for normalization via the RMS option. You can scan levels to see what the average decibel and peak decibel levels are.

We set voice for the highest levels, because voice is the most important sound in the game that needs to be heard clearly. This is a standard rule in film audio as well. Next, we set levels for sound effects as the second-highest category of normalized sounds, at –18 dB to –13 dB RMS. Sounds are almost as important as voice in terms of being heard clearly by the player, as they are vital for player feedback. Music is usually normalized at the lowest levels (the levels used on Deus Ex: Invisible War were –25 dB to –18 dB RMS or so), but these can vary depending on how they are used. If you want to use a piece of music to frighten the player with a quick staccato cacophony of instruments, for example, the normalization should be much higher than –25 dB RMS. Figure 4.7 shows Sound Forge's “average dB RMS” function at work.

3D propagation and falloff curves

While you can align your sounds perfectly for volume based on straight playback, your sound requirements change in a game in which distance is a factor—a common element in games with a first-person perspective. As if changing in distance relative to the listener, a sound's volume will change accordingly. Falloff curves are important to consider here: They describe how an object's sound will change as its distance from the player changes.

Say you have a sound for which you want to gain a certain amount of volume until it reaches its maximum (the average db RMS range discussed above). This could mean that some softer sounds will fall off too quickly if they're not as loud as others initially. For example, if you need footsteps to be audible while guards are shooting pistols, the pistol sound might be jacked up quite a bit while the footsteps get lost in the mix. You can change your falloff curve to reflect this so that the pistol sounds reach their maximum levels sooner than the footstep sounds do based on distance. Check out the chart in Figure 4.8. The curves represent exponential curves and linear curves, all of which are adjustable.

Figure 4.8. Distance falloff curves, from linear to exponential. An exponential curve is more commonly used, as it represents a smoother volume change to the average listener. For easier mixing later on, keep these curves adjustable based on either sound groups or even individual sounds.

You can use another option in place of or in conjunction with the falloff curve method: sound radii. In certain game engines such as Unreal, the developer has control over the radius in which a sound can be heard. Some engines have two radii, one representing the point at which the sound volume drops to zero, and the other representing the point at which the sound reaches maximum volume (the point at which it is normalized) in relation to the player. Increasing the radius for maximum volume would make the footsteps more audible, as they would hold their volume for a longer period of time depending on the player's movement in combat.

  • Creative Edge
  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint