Sound5 min read

A dialogue mix you can follow on a phone

Bring a soft voice, a busy room and a music bed together without losing the words that carry the story.

Record yourself saying, “Fold the blue strip over the green one.” Put a lively song underneath. On a phone speaker, being able to hear a voice is an easy test to pass. Being able to follow the instruction is a more revealing one. A listener who catches “fold the strip” has missed the two words that determine what their hands should do. Circle those words on your script. They give you a concrete way to judge the mix.

Build a speech-only version

Mute the music and effects. Choose a sentence delivered at the speaker's ordinary level and listen at a comfortable, repeatable playback volume. Work outwards from that sentence. When a guest leans away from the microphone for an entire answer, lift that clip a little and listen across the join. When only the last word drops, adjust the local phrase. Changing the entire track for one quiet word will also lift the breaths and chair squeaks beside it.

Listen to the original recording if a word already sounds fuzzy. The camera microphone may have a clearer copy of a syllable covered by clothing against a lapel microphone. Compare candidates at a similar perceived level to keep loudness from swaying the choice. Use a short crossfade where the room sound changes, then hear the replacement in the sentence. A repaired word that draws attention to the microphone has further work to do.

Choose where to spend the music

Draw three marks on a sheet of paper. Put one where the viewer sees the finished paper model, one where the first instruction begins and one where the last fold clicks into place. Listen through the music and find a suitable musical event for one of those marks. A change of chord beneath the reveal offers something to build towards. A busy guitar phrase beneath a spoken colour name gives the ear two intricate things to follow at once.

Try a section with fewer instruments under the instructions. If the music comes with licensed stems, audition the rhythm track alone and bring the melody back for the reveal. With a single stereo file, try another part of the recording or another piece with a simpler arrangement. That choice can recover speech detail with a smaller change in level. Keep the exact music credit and licence details with the project so another editor can find the same source.

Draw the volume curve across a whole thought

Start the music dip just before the speaker takes the floor and leave it down through a short hesitation. Raise the bed when the hands take over and there is something to watch quietly. In Premiere, tagging dialogue and music lets Essential Sound generate ducking keyframes. Listen to where those keyframes land before spending time fine-tuning their depth. A keyframe on every breath can make a patient demonstration sound as if someone is repeatedly reaching for the volume knob.

Play a sentence containing a deliberate pause. Pull the middle of the dip flat and compare it with a version where the music rises into the gap. Ask which version makes you expect the next instruction. In a tutorial, the flatter curve usually leaves room for the speaker to continue. In a reveal, a rise can give the viewer permission to study the image. Keep a copy of any hand-adjusted automation before regenerating ducking keyframes.

Let the room continue underneath an edit

Find a stretch of the same room with the participants still and the microphones in position. Listen for identifiable sounds such as a horn or a cough before repeating a section. Under headphones, a horn that returns every few seconds soon gives away the loop. Slide two pieces beneath the dialogue join and overlap them gently if you need more time. Lift only enough ambience to connect the surrounding takes.

For the fold itself, choose one close sound you recognise. The scrape of thick paper against the table can carry more information than a general whoosh. Sync the start to the hand movement, then play it beside the dialogue. If the scrape obscures “green,” move it to a visible adjustment just after the word, shorten its tail or take it out. Let the action on screen decide whether the change is believable.

Measure the deliverable you intend to send

Read the client's audio specification and check the complete mix against the stated loudness and peak limits. A loudness meter's integrated reading describes the programme across a full pass. A short-term reading can reveal why an individual passage feels louder than its neighbours. Reset the meter before running the whole sequence. Comparing a measurement taken over the final minute with a target for a full programme can send the mix in the wrong direction.

Export a test and listen through headphones, a phone and, if available, a single speaker. Keep the volume fixed during each pass. Write down the first word you lose, any jump that makes you reach for the controls and the time it occurs. Invite a listener who has never seen the sequence to follow the folding instruction without captions. When they pick the right strip and finish the fold, turn the captions back on and check that the labels agree with what they heard.

Sources