Said Again in a Booth

A film set is a loud place. Generators, wind machines, traffic, aircraft overhead, crew breathing, and the camera itself, which on large formats is a substantial piece of machinery running a foot from the actor's face. The production sound recordist's job is to get a usable voice out of that, and often it cannot be done.

The remedy is ADR. The actor stands in a studio, watches the shot loop, and speaks the line again in sync with their own lip movements. It is used for noise, for intelligibility, for lines rewritten after the shoot, for censorship, and sometimes for a performance a director has decided to change in the edit.

It has a cost that is not obvious until you hear it done badly. A recording booth is deliberately dead, with no reflections at all, so a line brought back from one has no room in it. If the scene takes place in a stone corridor, the engineer has to put the stone corridor back, matching reverb, distance and the colour of the original microphone, and if that work is imperfect the line floats slightly free of the picture. Audiences rarely identify what is wrong. They simply find the moment less convincing.

Which brings us to the most litigated sound question of the last decade.

Christopher Nolan's films attract a persistent complaint that the dialogue is hard to hear, and his explanation is partly this: he does not like ADR. He has said he prefers the performance given in the moment to an actor revoicing it later, and acknowledges that it is a choice people disagree with. His cameras are loud, so the production track needs cleaning, and software that strips camera noise is what makes intimate scenes shootable on that equipment at all.

That accounts for some of it. The rest is mixing, and here the issue is bigger than one director.

A theatrical mix is built for a specific room: a quiet auditorium, a calibrated system, a dedicated centre speaker carrying dialogue, and an audience with no volume control. In that room you can place a whisper far below an explosion and both will land, and the distance between them is an expressive tool. The quiet line before the blast is quiet on purpose.

Now move that mix into a living room. A television or a stereo soundbar has no centre channel, so a five or seven channel mix is folded down to two, which means the dialogue is arithmetically summed in with the music and the effects instead of arriving on its own speaker. The room is not quiet. The screen is a metre wide. The same decision that read as dynamic range now reads as mumbling.

Two consequences followed, and both are now normal. Subtitles stopped being an accessibility feature and became the default for a large share of viewers, and streaming releases increasingly ship a separate near-field mix with the dialogue brought forward.

It is tempting to award this argument to one side. Resist it slightly.

A filmmaker mixing for a cinema is mixing for the version of their film that was paid for and projected, and treating the home as the reference would flatten the theatrical experience for everybody who goes. A viewer with a soundbar is not wrong to expect to hear the words. The honest description is that there are now two reference rooms with incompatible requirements, and that a film which commits fully to one will disappoint in the other. The industry's answer, two mixes, is unglamorous and roughly correct.