Music bed
Also known as: underscore, bed
Music placed underneath other audio to support it rather than lead, sitting low enough that speech and real sound stay in front.
Updated
What it means
A music bed is a track doing a supporting job. It fills the space under narration or camera audio, giving a sequence continuity without asking for attention of its own.
That is a different role from a soundtrack carrying an action edit. A bed is chosen for being unobtrusive, which usually means less rhythmic definition and fewer events to cut against.
The distinction shows in the mix. A bed sits well below the foreground, while a Soundtrack that drives a highlight sits on top and the rest of the audio arranges itself around it.
How it works in practice
Level is the whole craft. A bed should be audible when attended to and unnoticed otherwise, which in practice means setting it considerably quieter than instinct suggests and then checking on small speakers.
Automatic level reduction keeps it honest under speech. Audio ducking drops the bed whenever words appear and restores it afterwards, so no fader has to be ridden by hand while the video plays.
Instrumentation matters more than tempo. Sustained textures sit under Dialogue comfortably, while anything with a strong transient competes for the same space as consonants and makes words harder to follow.
What to watch out for
A bed too loud is the most common fault in talking footage. The words are technically present and the viewer stops following them, which reads as the video being boring.
Beds with vocals almost never work underneath speech. Two voices at once is one too many, and a listener tries to parse both rather than treating either as background, which is tiring almost immediately.
Looping a short bed badly is audible even at low level. A seam every eight bars nags at a listener, which is why a Seamless loop matters more here than in a foreground track.
How RawClip handles it
Where a clip carries speech worth keeping, the music drops to a supporting level underneath it and comes back up afterwards without anybody riding a fader. The rest of the time the track leads, because camera audio from a moving mount rarely has anything to say.