Balancing a Reporter's Voice Against Music: Notes From Owen's Project

The report that kept getting quieter

Owen runs a small production outfit and had cut a ten-minute recorded report for a regional client. He sent it to us during a quiet week in mid autumn with a short note: the narration seemed to disappear whenever the music came in, and when he turned the music down, the transitions sounded oddly empty.

I listened on two sets of headphones and a small pair of desktop speakers before touching anything. The problem was not one fault but three small ones stacked together. The narrator moved closer to and further from the microphone between sessions. The music bed was mastered much louder than the speech. And an interview clip in the middle sat well below everything around it.

Owen had authorised recordings and an approved script, which made the job straightforward to begin. Everything that follows started from that listening pass rather than from any plugin or setting.

Listening before adjusting anything

Level balancing begins with a decision about what the listener must always hear. In Owen's report, that was the narration. Music and effects were there to support it, never to compete with it. I wrote that down at the top of my notes so that every later choice could be tested against it.

I then marked each spot where the voice felt uncertain. There were six: two music entries, one music exit, the interview clip, and two places where a sound effect landed on top of a stressed word. Marking them first meant I fixed problems rather than chasing a general impression of loudness.

One habit I teach on the course is to listen at a modest volume, not a loud one. A mix that holds together quietly will usually hold together on a laptop speaker, in a car or on a phone, which is where much of Owen's audience would hear it.

• Decide which element must stay clear at all times
• Mark every spot where the voice feels uncertain
• Check at a low listening volume as well as a normal one

Evening out the narration first

Before touching the music, I worked on the voice itself. The narrator's two sessions differed noticeably in level, so I matched the clips by ear, adjusting each section until the spoken delivery felt like one continuous performance. Only then did I use gentle compression to reduce the gap between his quietest words and his emphasised ones.

The aim was restraint. Heavy compression flattens a voice and brings up room tone and breaths, which creates new problems. I kept the settings light and listened for any pumping on the ends of phrases.

Once the narration sat steadily, many of the apparent music problems eased on their own. That is a common pattern: a voice that wanders in level makes the whole mix seem unbalanced, and fixing it first saves a lot of unnecessary work later.

Making room for music and the interview clip

The music bed needed two things. First, it came down overall so that it sat clearly beneath the speech. Second, I automated it to dip slightly under each spoken passage and rise again in the gaps, with fades long enough that the movement was felt rather than noticed.

The interview clip was the opposite problem. It was quiet and slightly dull compared with the studio narration, so I raised it and added a small amount of clarity in the upper midrange. I did not try to make it match the narration exactly. A field recording that sounds like a field recording is believable; one forced to sound like a studio rarely is.

Sound effects were trimmed so that none landed on a stressed word. Two were moved by a few frames, and one was simply lowered. Small moves like these did more than any broad change to the master level.

• Dip music gently under speech, with slow fades
• Lift weak interview clips without forcing a studio sound
• Move or lower effects that collide with stressed words

The review version and one round of feedback

I sent Owen a review version with the six marked spots listed in a short note, so he knew where to listen. He came back with one comment: the music exit near the end felt slightly abrupt. That was a fair point, and well within the agreed scope for a single round of revisions.

I lengthened the fade by a second and a half and resent the file. Owen approved it the same afternoon. Keeping feedback attached to specific moments, rather than general feelings, made the exchange short and calm for both of us.

Final delivery followed the usual order of checks: confirm the loudness suits where the report will be played, name the files clearly, and keep an archive copy beside the delivery file. Nothing unusual was needed, and that was rather the point.

What Owen's case shows about balancing

The lessons here are small and repeatable. Decide what the listener must hear, fix the voice before the music, treat each recording on its own terms, and keep adjustments modest enough that nobody notices them. A balanced mix is mostly one in which nothing draws attention to itself.

On the course, learners practise this on authorised recordings with an approved script, starting with the narration and working outward. By the time they reach music and effects, the decisions are usually simple.