Producing Great Sound for Digital Video

By Jay Rose

The graphics in the book are bigger and higher resolution than these on-line samples.

Here are a few tiny excerpts from the text:

From Chapter 1: How Sound Works


"Architecture is frozen music."

When Friedrich von Schelling wrote that, he was being poetic. The scientific truth is if there were such a thing as frozen music, it would be pressure. Usually air pressure; but it can also be water pressure, wood pressure, or pressure in anything else that conducts sound. Just pressure. Not the kind of pressure you feel when deadlines approach, but pressure that compresses molecules together. These molecules can be of almost anything: sound usually travels in air, but the same principles hold when it's traveling in water or across a solid.

If the pressure keeps changing repeatedly within certain speed ranges, you've got sound. Something -- a drum skin, human vocal cords, the cone of a loudspeaker, or anything else that makes noise -- starts vibrating back and forth. As its surface moves towards us, it squeezes air molecules together. As it moves away from us, it creates a very slight vacuum that pulls the molecules apart.

If we could freeze sound and see the individual molecules, they'd look like Figure 1.

Figure 1: If we could see sound, it would look like this. Air molecules are squeezed together and pulled apart by the vibrations of the tuning fork.

Think of Figure 1 as a snapshot of a single moment in the life of a sound. Air molecules are represented as tiny black dots . . . and as we enlarge sections of the picture, we can see individual ones.

The life of a sound

The tuning fork vibrates back and forth. When its surface moves towards the air molecules next to it, it squeezes them together. Those compressed molecules push against the ones a little farther from the tuning fork, and that squeezes the farther molecules together. The farther molecules now push against ones even farther, and so on: as the squeezing spreads out to successive layers of molecules, the pressure spreads out.

Air molecules, like everything else in the physical universe, take time to move from one place to another. So even while the pressure is spreading outward, the tuning fork -- which is vibrating back and forth -- may start moving back in the other direction. The air molecules next to the fork rush back in to fill the space where it was, pulling them a little further apart than normal. This very slight vacuum -- engineers call it rarefaction -- pulls on the next layer of molecules a little farther from the tuning fork, spacing them apart. And the process repeats to successive layers.

From Chapter 5: Pre-production

On-set playback

Some sequences, particularly musical ones, require that talent work with a previously recorded track. If you don't intend to record dialog or effects during these sequences, the audio setup is simple: all you need is a stable playback medium such as DAT or Minidisk (the speed of analog audio cassettes drifts too much to be useful for playback), speakers close enough to the talent that they won't be confused by room echoes, and a feed from the playback device to your recorder for sync reference.

But these sequences can be more effective if you're also recording natural sounds, dialog, or the characters singing. This kind of shot requires preproduction planning, and you'll need an experienced production sound recordist to help you sort out the options:

Shoot now, record later

Some locations have too many problems to allow successful dialog recording, and a dialog replacement session may be the only way to get an acceptable soundtrack. ADR (automatic dialog replacement, but it isn't really automatic at all), also known as looping, is an accepted part of Hollywood filmmaking. So are budgets that reach a hundred million dollars.

ADR is expensive, exacting, annoys the actors, and seldom gives you a wonderful soundtrack. It should only be considered as a last resort in video production; often, compromising slightly on the choice of location will yield a better overall project. If you must use ADR, it helps to plan ahead:

The decision to use ADR doesn't absolve you of the responsibility to record sound in the field. Your actors will need a guide track to work efficiently, and some of the original dialog or effects may be salvageable.

The actual techniques for ADR are discussed in Chapter 8.

From Chapter 7: Recording Dialog

Operating the boom

It's no problem to work a mic boom (assuming you have enough stamina to hold your arms over your head for long periods of time), but it's not necessarily intuitive.


After you extend each boom section as far as it'll go, collapse the joint slightly so the pieces overlap a couple of inches. This will make it sturdier and less likely to creak at the joints. Figure 1 shows the wrong (top) and right (bottom) way to extend. It's better to extend three sections partway than two sections all the way.

Figure 1: If a boom section is extended all the way (top), it won't be as strong as if it overlaps (bottom).

The clutch that locks the sections of a boom when they're extended can wear out and become difficult to use. A layer of plumber's Teflon tape around the threads usually fixes things.

Sometimes sections of a collapsed fiber boom can get stuck together, particularly if it was moist when stored. Gentle heating can usually get things moving again.

Overhead booming

Hold the boom with your arms straight up in an H position (Figure 2), rather than out like a Y (Figure 3). This way the weight is being supported by the bones in your arm, not by your muscles.

Figure 2: It's easier to hold the boom with your arms straight up . . .

Figure 3: . . . not out to the sides.

It may be tempting to hold the boom like a flagpole (Figure 4), but this gives you less control over how the mic is aimed. Since the boom itself will be at an angle, it's likely that it'll cross into a corner of the shot.

Figure 4: Don't hold the boom like a flagpole.

If you can, hold the pole a foot or so in toward the middle rather than at the very end. This way part of the boom can act as a counterweight, and you won't be stressing your wrists as much.

Cable management

The mic cable can make noise as you move the boom, by banging around inside the boom pole. This gets conducted up the cable and into the mic.

Isolate the microphone end by making a small loop in the cable and securing it with a piece of tape. The loop should be just big enough to keep from slipping into the boom.

Then, loop the mic cable around your finger where you're holding the boom, and pull slightly to keep the cable taut.

If you're using a boom where the cable has to run outside, tape the microphone end of the cable tightly to the boom. Wrap the cable around the boom a few times on its way down, and grab the cable along with the boom.

As the boom is swung around, a loose XLR connector may rattle and create electrical or mechanical noise. To be safe, secure the connector to the mic by wrapping it with tape.

From Chapter 12: Working with Music

Cutting by counting

Once you've learned to count accurately with the music, the editing part is simple. All you have to do is match the numbers. Start by loading track 24 of the book's CD into your editing system. This is one of a number of pieces written specifically for this book by Doug Wood, president of the Omnimusic Library, and it's designed to be typical of the kind of straight corporate or documentary theme you'd find in most libraries. It counts in 4.

If you're using a desktop NLE, follow the steps below. If you've got a different kind of system, the procedure will be only slightly different. Read along through these instructions, and we'll cover the differences later.

When you're done, you should have a bunch of marks looking something like the gray flags in Figure 9.

Figure 9: Tap hard enough to make a mark each time you count the number 1.

Copy the clip to a timeline, and you're ready to edit. If you want to shorten the piece, cut out the space between two marks and butt the pieces together. In Figure 10, we cut from Marker 2 to Marker 4, pulling out about three seconds. (Because of the way chords flow in most music, it usually sounds better to cut from an even number to another even one, or from an odd number to another odd one.) If we wanted to extend the music, we could have taken that stretch between Marker 2 and Marker 4 and laid it into the timeline twice.

Figure 10: Shortening a cue by cutting from one numbered mark to another

This is a very short edit -- there's only about three seconds between the markers -- but the technique works for edits of any length. If you need to edit longer stretches of music, make the marks farther apart. . . say, every fourth "one".

Editing off the marks

As you listen through the sample corporate piece on the CD, notice how the horn melody begins around 12 seconds in. It actually starts on a "4," one beat ahead of the downbeat. But you'll still use those markers you made to cut the song accurately, and start that melody anywhere you want.

Figure 11: An edit can be before or after a marker, but the markers still have to line up.

In Figure 11, we built the edit on two tracks and added a quick fade up to make the entrance of the horn melody smoother. But you can do a similar edit on a single track, making a butt cut from one marker to the other, and then rolling the edit point forward to pick up the melody. The important thing is that, no matter where you make the edit, you still preserve that constant heartbeat -- the length of time from one downbeat to the next has to stay the same.

©1999 Jay Rose
First posted 22Nov99