The Meters, Monitors and Methods Behind Our Level Balancing

Why we start with a meter, not a fader

When a producer sends us a programme segment, the first thing we reach for is a loudness meter rather than a fader. Ears tire quickly, and after twenty minutes of listening, a voice that was too quiet at the start can seem perfectly fine. A meter does not tire. It tells us the integrated loudness of the whole piece, the loudest short bursts and the range between the quietest and loudest passages.

We use that reading as a starting position, not as a verdict. A number can show that an interview sits several units below a voiceover, but it cannot tell us whether the interviewee sounds distant for another reason, such as room echo or microphone distance. The meter points us to where to listen, and then we listen.

Learners on the course are asked to measure before they touch anything. It feels slow for the first few exercises and then becomes habit.

• Measure the whole segment before any adjustment
• Note the loudest and quietest passages
• Treat the reading as a prompt to listen

The monitoring setup we trust

Balancing is only as good as what you hear it through. We work on a pair of nearfield monitors in a treated room, and we check the same mix on closed-back headphones. Most of our clients' audiences will listen on laptop speakers, phone speakers or ordinary headphones, so we also keep a small speaker and a cheap pair of earbuds on the desk.

Switching between these is the quickest honest test we know. A music bed that feels tidy on the monitors may swallow consonants on a phone speaker. A voice that seems warm and full in headphones may turn boxy on a laptop. None of this requires expensive equipment, and we encourage learners to build a similar routine with what they already own.

We also keep our listening level fixed. If the monitor volume wanders from session to session, every decision about balance wanders with it.

• Nearfield monitors plus closed-back headphones
• A small speaker and earbuds as reality checks
• One fixed listening level for every session

Clip gain before compression

Our method follows a set order, and the first pass is clip gain. Before any compressor is inserted, we go through the dialogue and lift or lower individual phrases so that the spoken level is already reasonably even. A speaker who leans away from the microphone for one sentence is fixed with a few decibels on that sentence alone, which no compressor can do as cleanly.

This matters because a compressor given uneven material will react to the loudest syllables and squash everything around them, raising the noise floor and flattening the delivery. Starting with clip gain means the compressor has less to do, so it can be set gently and the voice keeps its natural movement.

It is slow, careful work, and it is the part learners most often want to skip. It is also the part that most changes how a finished segment feels.

Setting speech against music and effects

Once the dialogue is steady, we bring in music and effects. Speech is the anchor: it is balanced first and stays where it is, and everything else is fitted around it. A music bed under narration usually needs to sit well below the voice, and we often trim a narrow band in the music where the voice carries most of its clarity, so the two do not compete.

Where music rises between sentences, we write the level changes by hand with automation rather than relying on automatic ducking alone. Automatic ducking tends to pump, with the music audibly dipping and returning. A hand-drawn curve follows the rhythm of the speech and goes unnoticed, which is exactly what we want.

Effects get the same treatment. A door close or a title sting should be felt as part of the scene, not announced as an event.

• Dialogue first, then music, then effects
• Trim the music where the voice needs room
• Draw level changes by hand where it matters

Matching speakers across a whole piece

Interview films and recorded reports often combine several voices recorded in different places. A presenter in a quiet studio and a guest in a busy office will not match by level alone, because their tone, room and noise differ. We match loudness first, then listen for tonal differences and adjust gently so that the cut from one voice to another does not make the listener reach for the volume control.

We avoid chasing a perfect match. Two people never sound identical, and over-processing one voice to resemble another usually removes what makes it believable. The aim is that nobody notices the join.

A useful habit is to listen to every transition with your eyes closed and ask whether anything changed besides the person speaking.

A final pass before review

Before any version leaves our desk, we play the whole piece from start to finish at a normal listening level, without stopping to adjust. Stopping hides problems, because you fix the moment in front of you and forget the overall shape. Listening straight through reveals the places where the balance drifts, where a section feels louder than it should, or where an ending is too abrupt.

We then check the final loudness reading against the delivery target agreed with the producer and note it with the review version, so there is a clear record of what was aimed for and what was achieved.

Only then is the review version prepared. Anything we notice at this stage is fixed first, so that feedback from the producer can concentrate on creative choices rather than avoidable faults.