We have a few weeks' breathing room before the Edinburgh Film Festival deadline, so we're going to be tweaking audio. There are three places in the soundtrack where (we've determined) some people have difficulty understanding a few words.

Unlike many short films (and independent features, for that matter), CREDO doesn't have the problem of poorly-recorded, muddy dialogue. Our dialogue track is gorgeous and clean. We don't face The Low-Budget Filmmaker's Challenge; we face The Rock Engineer's Challenge: Because it's God, and he sings opera, and there are big thunder sounds and orchestra/choir/percussion hits accompanying his singing, it's impossible for me to know where the line between big and powerful and buried to the point of unintelligibility lies—at least, not until after a few test screenings on various home entertainment setups, using various sets of human ears.

Intelligibility is largely psychological. It's not entirely about ears and soundwaves. For instance, one of these points occurs during a very wide shot, in which God is almost a dot in the middle of the screen. If it were a closeup, I don't think the issue would exist—because we'd see his lips move, and that would contribute a powerful psychological cue to our understanding of what he's singing.

Along the same lines, if the engineer (in this case, me) already knows the lyrics and the performance by heart, there's no way not to understand them. I can hear the words perfectly in all three of these spots.

In two cases, fixing it is simply a matter of turning down the volume (slightly and briefly) on everything except God. (For you geeks, it'll probably be something in the neighborhood of -3 dB for about two seconds.) In the third—the only spoken scene—we'll be replacing the dialogue with a studio track, either this weekend or early next week. One of the nice things about working with singer/actors is how accurately they can duplicate their performances in post, timingwise.

The problem in this last case was the director's fault (also me). The original plan for this shot was that we would do dialogue replacement in post. It gives a track a subtly unnatural quality, which was exactly what I wanted. Based on this plan, I told the location audio guys not to bother moving their whole setup just for this single exterior shot.

And then, of course, once I saw Larry's performance during editing, I decided to cancel the dialogue replacement and use the location sound—which was recorded from fifteen feet away, with wind and traffic noise, using the camera's microphone. Lots of EQ later, I thought I'd brought the vocal sufficiently to the foreground. The broadband noise behind him also increased, but I thought the procedure had been successful. Judging from our test screenings, apparently not.

That's the only negative comment we've received from anyone—and interestingly, everyone says the same thing:

"There were a couple places where I couldn't understand him, but that might just be my system."

The other thing everybody says is:

"I just love him [Larry] as God."

Intelligibility in a massive sound mix? That, I can fix.

Without a great performance, though, it wouldn't matter.


[ Official CREDO website | Notify me when new entries are posted ]

Your comments and questions are welcome: