Sound types, dialogue repair and Match loudness per type

Sound in a film falls into a few broad families, and each is treated differently. Dialogue needs to be clear and steady. Music sits underneath. Effects punctuate. Ambience fills the space so silence doesn't feel dead. Video by Zaweba lets you tag every sound clip with its type, and then uses those tags to help you repair and balance it.

Tagging clips by type

Each clip can be tagged as dialogue, music, effects or ambience. You don't have to start from nothing: the app makes a first guess at the type from the clip's level envelope, the way its loudness rises and falls over time. Speech has a typical stop-start pattern, music tends to be more continuous, and short sharp spikes suggest effects.

Treat the guess as a head start, not a verdict. Look down the list and correct anything that's wrong, such as a quiet music cue taken for ambience or a shouted line taken for an effect.

The dialogue repair group

Once speech is tagged, the dialogue repair group gathers the tools for fixing it in one place. Dialogue is where repair matters most, because faults in speech are noticed instantly and make people stop listening. The tools in the sound effects library for this kind of work, such as noise reduction, de-essing, de-reverb and click and clipping repair, are the ones to reach for here.

Work on dialogue before anything else in the sound pass. Clean, even speech is the foundation the music and effects are balanced against.

Match loudness per type

Different recordings come in at different levels. One interview recorded close to the microphone is loud; another from across the room is quiet. Match loudness per type evens out clips of the same type so they sit at a similar loudness. All your dialogue lands in the same range, all your music in its own range, and so on.

That's very different from making everything the same level. Music under speech should stay below the dialogue, and ambience below that. Matching per type keeps each family consistent within itself while leaving room between them.

Where this fits in a sound pass

A practical order looks like this. Tag every clip, check the guesses, and fix wrong ones. Repair the dialogue. Match loudness per type. Then mix the families against each other in the sound mixer, and keep an eye on the loudness meter on the master.

Tags pair well with tracks and submixes. If you also route each type to its own submix, the mixer gives you one fader per family, and the structure of your project mirrors the way you think about the sound.

Things to watch for

Mixed clips are the tricky case. A clip recorded on location might have dialogue and loud background music in one file. Tag it by what matters most, usually the dialogue, and handle the rest by hand.

Recheck after big edits. New clips arrive untagged until you or the guess sorts them. A quick look before the final mix stops one untagged line from dodging the matching. Done this way, Video by Zaweba keeps a busy soundtrack orderly without a lot of manual level setting.