SKIP: Directors ask for sounds generally in 2 ways: they are very articulate when looking for an important sound that will stand out; but for background or minor sounds, it's usually something like, "I can't hear that car in the background. We should hear it." The one we pick may not be the one the director had in mind, but usually they don't really care, as long as the atmosphere is appropriate.

TOM: I worked on a film that had an ice skating sequence, and the director rejected the sound that we proposed, saying "It's got to be more sharp, hard, gritty." What we had was more of a swish, whereas what he wanted was more of a scrape. And sometimes that's hard to interpret.

SKIP: One man's swish is another man's scrape...

TOM: Yeah, we kinda have to guess.... I can do a certain amount with equalization and things like that, but there comes a point where we just have to get rid of the sound, and get another one from Skip.

What are some of the ways you use to personalize and enliven a soundtrack?

SKIP: A really easy thing to do, especially with Foley sounds, is to make small sounds really loud. It forces the listener into a weird perspective where they wouldn't ordinarily hear that. Like breathing -- in normal circumstances you don't hear that, but if it's really loud, the audience notices that. Like what we did in Silence of the Lambs, in that scene where Jodie Foster is being stalked by that guy wearing night vision goggles. All you hear is these weird background sounds, and breathing. And her footsteps.

Plus we made some weird background sounds. We took wolves, and slowed them down tremendously, so that all you had were artifacts of the process, and that became part of the sound. That was just wild fooling around -- there was no particular goal, we just kept going and going, with the gear; the gear had as much to do with that sound as we did. But mostly it's just normal sounds; only occasionally do we get to do things like that.

TOM: This film [Clockers] is pretty straightforward. Except we put a lot of sirens in the background -- we're not sure if [Spike Lee] is going to go for that. But we just put in sirens, as ambience, there all the time. They get louder and softer. I tried to play those low enough so that the director doesn't say, "Whoa, what is that?" I've tried to keep them down low enough so that you don't really notice them, but cumulatively, it has an effect over the course of the film. But also, the music might drown it out -- it probably will, in a lot of cases. But they'll still be there. And my kind of Zen way of thinking about it is, if it's in there, even if it's drowned out in places, it will still affect the experience

How do you create unreal, otherworldly sounds?

SKIP: You just have to go with your instincts. Like in Star Wars: Ben Burtt, who did that brilliantly, made up a lot of the sounds in there. Like the Wookie was made up mostly of sounds of camels. The spaceships -- one of those is made mostly from camel sounds. The light sabre -- Ben has this old TV. He took a microphone, and put it in a metal tube. The TV made a lot of noise. So when he would move the mic close to the TV, you would get that sound. And it was brilliant -- you get exactly the sound you want. Now, it's not difficult to see how you could make that -- a kind of noisy, electronic sound. You could see how he got that. I never got that . I do it all the time -- you go to movies all the time, and you go, "I never would have thought of that. I should've just gone to law school."

Do you use samplers?

SKIP: We use Synclaviers, but a sampler is similar -- basically just a set of switches with intervals between them -- that's all a keyboard is. It is more sophisticated than that, but that's the essence of it. In the Synclavier itself, it has the sequencing software, so that you can add to or take away from what you've recorded. But you can use any sampler and sequencer, and whatever controller that sampler comes with, and you can do that. The essence of the Synclavier is that you have the sampler, you have the variable pitch, because you have all these intervals between switches [keys]. So one of your parameters could be variable pitch; what that does is just changes the sample rate. So if your sample is 50Hz, and you play it at 25Hz, it's going to be one octave lower, and it will be twice as long. That's a fundamental type of manipulation.

Another modifier might be the traditional ADSR -- attack, decay, sustain and release -- fairly widely-known in synthesizers. That's what the Synclavier does, pretty much. When you push the key down, there's the attack, you hold it for sustain, you release, and the decay is whatever you've set -- how long the sound takes to fade.

The Synclavier is just a digital recorder -- a sampler. What we do is record stuff, from somewhere -- from a library, from DAT, from a microphone. Then you use this device to manipulate the recording. Fundamentally, all we really do is manipulate recordings.

What would be an ideal interface to a database of sounds?

SKIP: I could imagine if you had a mixing bowl, with a sort of batter in it -- your sound -- that you were making. And around the bowl you could have all different measuring cups representing sounds you were putting into the bowl. And you could put the mouse over one and click "plus" and it would get bigger....

Everyone else uses a pretty much standard interface. All the editing programs that we use all have it. Everyone pretty much, instead of creating a new interface, they say, "We want our interface to look like all the other interfaces, because we know how to use those. The last thing we want to do is make some kind of new thing." But my daughter and my son, they play on my Mac, and they don't know what I'm doing. They could much more easily relate to the mixing bowl metaphor.

TOM: It's hard to represent, say, a dog bark with a shape. But you could represent the envelope as, say, a curve within a shape -- showing the way a sound grows and decays.

SKIP: You could try to break down the sounds into groups. It's fairly common. There are pretty fundamental types -- voices, animals, atmospheres, footsteps....

TOM: Certainly those things could be represented. Like you have a picture of a car. And you push that, it opens up, and you have cars, trucks, buses.... Once you get down to describing it, you could say, "Traffic with a lot of horns," "High-angle traffic," "Close-up traffic." That's what I get on the cue sheets. And I'm familiar enough to know what the difference will be. Then what I do is just try to blend them in some way, so it's appropriate for the scene.

SKIP: There's also another problem, and that is, as a user -- let's say you have a relatively sophisticated database... you could, say, type in "traffic." Presumably, there would be several places where you would get a positive "hit" for that word. I don't want to toggle through a bunch of sounds, or scroll through stuff. I want to type in "Midtown traffic car-by" or something like that.

TOM: Basically, the situation you will have is, you'll have this library, and someone will have something in mind that they want, and they're going to want to have choices. So you're going to want to give them a number of choices, then have them click a mouse and hear it. Because that's what it all comes down to -- you want to hear it play.

SKIP: Concievably, you could take all the sounds that are in a category -- let's say animals -- and you could compile a reference recording of all the sounds -- that would be one recording. And you could hold a magnifying glass icon, as you move it, depending on where you are, you hear, say, dog barks and cats.... Then, you could have text associated -- now you're listening to these three sounds together. Or, move down the line, and you would hear a horse, a pig, a cow and a dog. And you would say, "Now you're listening to a farm atmosphere."

Concievably, if you broke your library down into ten or twelve groups, you could have a different thing for each group. So if you have all your traffic, maybe you move your mouse over the top, and it says "Midtown traffic," and as you move down, you have fewer cars, quieter traffic sounds. If you had water, you could go from a loud waterfall, to surf, to drops.... Maybe you have a dialog box that pops up and tells you what it is, as your dragging over and acutally hearing the sounds.

That's the way the Help bubble works on the Mac. Anything you put the cursor over, a bubble pops up and tell you what you're looking at. It's horrible, but, yeah, that works. Actually, there is a workstation that has that. Because it's infuriating when you have a graphical interface, but you have no idea what the graphics represent. You can look at a text field, but that's equally useless. What we have now, is a blue line called the "now line," that tells you where you are, and you can put the cursor on a part, and click, and all the text that's associated with that cue pops up in a box.

Whether it's icons or quick keys, you have to take the time to learn what that means. The Synclavier, on the other hand, for instance has a scissors icon, and you know exactly what that means.

TOM: I think you'll find that the type of graphics you'll need will be defined by the functions. The type of machines we use have way too many functions to graphically represent. They come with a thing called "Icon Pro" where you can design icons for specific functions, We don't use that because, just to think of an icon that will describe, say, a ripple into video black, what will that look like?

SKIP: The Synclavier is an example of multiple interfaces -- you can work on the keyboard, or you can work on the Mac. That's very powerful.

We're looking forward to the day when computer code -- clock cycles -- will replace time code.

Thanks to the SoundSphere group: Sacha Hare, Danny Scheman, Karthik Swaminathan, Alyssa White!