A strong breaking-news bed can give a bulletin urgency, pace, and a clear sonic identity. But once a presenter starts speaking, the mix has one job: make every word intelligible. That is the practical issue behind how to duck narration - although in most newsroom workflows, you are actually ducking the music bed when narration begins.
The distinction matters. If you lower the voice to make the music more prominent, you risk losing names, figures, locations, and qualifiers that carry the story. A proper duck keeps the narration stable and lets the music move out of the way, then return cleanly during pauses, transitions, and the outcue.
Start with the right bed, not heavy processing
Ducking is a mix tool, not a fix for a track that is too busy for the script. A bed with dense midrange synths, prominent guitars, aggressive percussion, or a lead melody can still compete with speech after you lower its fader. The problem is masking: the music and the voice are fighting for the same useful frequencies.
For voice-led reports, begin with a dialogue-friendly underscore version. These arrangements are built to support a presenter rather than claim the foreground. They commonly use less melodic movement, more controlled percussion, and space in the midrange where speech needs to sit.
This is particularly useful in short-form news video, where a producer may have only a few seconds to establish a story before the voiceover begins. A full headline theme can work well for an opening shot or title card, while the underscore takes over under the script. Using separate edits is cleaner than trying to force one loud, fully arranged track under an entire report.
Set the narration level before you duck the music
Do not automate the bed until the voice is edited, cleaned, and set at its intended delivery level. If narration gain changes later, every music move may need adjustment.
For broadcast and digital production, your final loudness target depends on the outlet, platform, and delivery specification. The important operational rule is consistent narration. The listener should not have to reach for the volume when a new clip, reporter track, or host line enters.
Once the voice is stable, bring in the music at a low starting point. There is no universal fader number because voices, arrangements, compression, and final loudness targets vary. As a working test, lower the bed until you can follow every word at ordinary monitoring volume, then lower it slightly more. News mixes often sound more balanced in a quiet edit suite than they do on a phone speaker, laptop, car radio, or television set with built-in speakers.
A useful reality check is to listen from outside your primary listening position. If the words remain clear but the bed still supplies pace and tone, you are close. If the music becomes the first thing you notice, it is probably too high for spoken copy.
How to duck narration with volume automation
Volume automation is the most controlled way to duck music under narration. It lets you shape the bed around the editorial rhythm instead of allowing a processor to react the same way to every word.
Start the bed at its full, intended level before the narration enters. Then place the first downward move just ahead of the first spoken syllable. This small lead-in prevents the music from overlapping the opening consonant, which is where a sentence can feel crowded.
Bring the bed down over a short ramp rather than an abrupt vertical move. For urgent headline reads, the fade may be quick. For a measured investigative report or business explainer, a slightly longer transition will sound more natural. Hold the reduced level for the duration of the speech, then raise the bed during a genuine pause, a visual transition, or after the final word.
Avoid bringing it back too early. The final word may be followed by a short breath, a station tag, or a natural tail in the voice recording. A music rise that lands on that tail can make the edit sound rushed.
Automation is especially effective when a narration track contains quoted clips, pauses for graphics, or alternating presenters. You can let the music lift between sections without making the entire report feel flat. It also gives you precise control over signature moments such as a stinger into a live hit or a bed rise at the close of a package.
Use sidechain ducking when the turnaround is tight
Sidechain compression, often called auto-ducking, lowers the music whenever it detects narration. Route the narration signal to the compressor on the music track, then set the compressor to reduce gain only when the voice is present.
This can save time on long explainers, podcast segments, and fast-turn social clips. It is also useful when a script is still changing and you do not want to redraw automation after every new pickup.
The trade-off is that sidechain ducking responds to sound, not editorial intent. It may pull the bed down for breaths, mouth noise, off-mic chatter, or a noisy interview recording. It can also rise in the middle of a sentence when the speaker pauses. That pumping effect is distracting in serious reporting.
Use moderate settings first. A ratio around 2:1 to 4:1, a short attack, and a release that returns the bed smoothly after phrases is usually more useful than aggressive gain reduction. Aim for a few decibels of reduction before reaching for extreme settings. If the compressor has to cut the music heavily to make the voice understandable, change the bed level, use a lighter underscore, or revise the arrangement of the edit.
For the cleanest result, combine both methods: sidechain ducking for the broad response to speech, then volume automation to control the opening, the closing, and any editorially important pauses.
Make room with EQ before cutting more level
Lowering the bed is not the only way to improve narration clarity. If music is masking speech, a small EQ adjustment can create space without making the track disappear.
Speech intelligibility often depends on the midrange, though the exact area varies by speaker and recording. If a bed has sustained pads, brass, guitars, or synth parts concentrated in that range, try a gentle, broad reduction on the music rather than a narrow and severe cut. The goal is not to hollow out the track. It is to stop the bed from competing with consonants and the leading edge of words.
A dynamic EQ can be useful here. Set it on the music track so the problem range dips only while the narrator speaks. This is more transparent than permanently removing the same frequencies from the bed, especially when the music needs to return to full impact between spoken sections.
Be cautious with EQ on the narration itself. Excessive high-end boost can make a voice harsh, while aggressive low-frequency cleanup can make it thin. Fix obvious rumble and room noise, then prioritize a natural, consistent read.
Match the duck to the format and the script
A 15-second social headline, a two-minute local package, and a thirty-minute news podcast should not be mixed the same way. Short headline content can tolerate more obvious music movement because the pace is compressed and the music is part of the format. A longer voice-led report benefits from steadier, less noticeable ducking.
The script also determines the approach. Financial figures, election results, emergency instructions, legal language, and names with unfamiliar pronunciation need more space around the narration. In these cases, the bed should support the gravity of the report without asking the audience to work harder to understand it.
Conversely, a short teaser with only a few words may allow a stronger bed and a more dramatic rise. The decision is editorial, not merely technical. Ask what the audience must retain after one listen.
Check the mix in the places viewers actually hear it
Before export, monitor the edit on more than studio headphones. Test speakers with limited bass response, a phone, and a modest laptop if those are relevant to your audience. Music that sounds controlled on headphones can mask speech on a small speaker, where the voice and bed are both pushed into a narrower range.
Also listen through the transitions. The most common failures are a bed that starts too loudly under the first word, a sidechain release that rises mid-sentence, and a music swell that covers the final line. These are small timing errors, but they make a finished bulletin feel less reliable.
When selecting music, choose packages that include timed edits, stingers, and dialogue-friendly underscore versions. Newsbed, for example, packages these practical cuts with each track so you can move from headline open to voice-led report without rebuilding the music for every duration.
The best duck is rarely the one a listener notices. It is the one that lets the report keep its momentum while every name, number, and essential line arrives clearly on the first pass.