Workflow guide
Music that lifts the video instead of drowning it
Background music sets pace and emotion, but the wrong track or a lazy mix buries dialogue and earns copyright claims. Good music editing means a licensable track, edits that land on picture beats, and voice that always wins. Here is the full pass from selection to final mix.
From the Mellius product team · Updated September 28, 2026
Conspiracy theorists, you may hate this
AI Director
›
Just type what you need
- ›Remove all silent parts
- ›Add bold kinetic captions
- ›Only have Bill Ackman speaking
- ›Cut to highlights only
- ›Add cinematic transitions
- ›Match cuts to the beat
The short answer
Choose a licensable track matching the video's energy, cut it to start under the hook and end on a downbeat, loop or extend cleanly for length, duck it six to twelve decibels under dialogue with slow fades, and export a loudness-checked master while keeping the license receipt with the project.
Licensed for the platform
The track's license covers every destination you publish to, including commercial and client work.
Voice always wins
Dialogue stays intelligible on phone speakers with music sitting clearly underneath, never beside it.
Edits land on beats
Music starts, changes, and ends on picture moments so the track feels scored rather than pasted on.
Pick a track that fits picture and paperwork
Match energy before genre: a calm explainer wants steady mid-tempo underscore with soft transients, while a launch cut wants driving percussion that survives phone speakers. Audition tracks under your actual voiceover, not solo, because half of all music picks fail the moment dialogue enters. Check the license for each platform — YouTube's Content ID system flags commercial tracks fast, and client work usually needs a commercial or business license rather than a creator subscription.
Keep stems or loopable sections in mind when choosing. Tracks with a clean intro, a steady body, and a resolved ending cut to picture far more easily than songs with vocals fighting your narrator. Common pitfall: falling in love with a track that has a singer in the same frequency range as your voiceover. Instrumental beds with restrained high mids almost always mix faster.
- Audition every candidate under the real voiceover.
- Prefer instrumentals when dialogue carries the video.
- Save the license receipt with the project file.
Cut to picture and duck under voice
Place the music start under the hook, not over the cold open's first word, and trim the ending to resolve on a downbeat with a half-second to one-second fade rather than a hard stop. For videos longer than the track, loop a steady middle section across matching downbeats instead of repeating the distinctive intro, which listeners notice immediately. Cut around scene changes: drop the bed out briefly before a big reveal, then bring it back to mark the new section.
Balance levels so dialogue peaks around negative six decibels and the bed sits six to twelve decibels below speech, then automate or sidechain the ducking with medium attack and slow release so the music breathes instead of pumping. Check the mix on a phone speaker at half volume — if you lose a single word, the bed is too loud. In Mellius you can ask for music under the voice at a named level and adjust it conversationally, with every change staying reversible on the timeline.
- End tracks on a downbeat with a short fade, never a hard cut.
- Duck six to twelve decibels under dialogue with slow release.
- Verify intelligibility on a phone speaker before export.
Compare the workflows
| Source | Best fit | Trade-off |
|---|---|---|
| Stock library subscription | Regular publishing with safe licensing | Popular tracks can sound familiar |
| Platform audio libraries | Free tracks cleared for that platform | License may not cover other platforms or clients |
| Commissioned or original music | Unique sound matched to picture | Costs more and takes longer to produce |
A practical sequence
Put it into practice
- 01
Select and license
Pick an instrumental-leaning track, audition it under dialogue, and confirm the license covers all destinations.
- 02
Edit to picture
Cut the start, loops, and ending to scene beats so the music feels scored to the visuals.
- 03
Mix and check
Duck the bed under voice, verify every word on a phone speaker, and file the license with the project.
Frequently asked questions
How loud should background music be under voice?
Start with the bed six to twelve decibels below dialogue and adjust by ear. Spoken tutorials and interviews sit nearer twelve decibels down, while energetic montages can ride closer to six. The phone-speaker test is decisive: if any word gets lost at half volume, the music is too loud.
Can I use any song I like as background music?
No. Commercial songs need a sync license, and uploading them without one risks muted audio, blocked videos, or copyright strikes. Use stock libraries, platform audio libraries for that platform, or commissioned tracks. Always keep the license receipt, especially for client work.
How do I loop a song that is shorter than my video?
Find a steady middle section and loop across matching downbeats so the rhythm continues unbroken. Avoid repeating distinctive intros or vocal hooks, which listeners catch immediately. Add a short crossfade at each loop point and vary the bed with small dips at scene changes.
Continue reading
Describe the edit. Keep the timeline.
Mellius turns conversational direction into cuts, captions, graphics, motion, color, and sound you can still inspect and change.
Start editing free