Keeping the original audio in a PMV

A pure PMV throws away the source audio and runs on music alone. But a lot of the best edits don't — they keep the original audio sitting under the track, so you still hear what's happening. It's most of the appeal of an audio-led edit, and it's the difference between watching a montage and being in the scene.

The problem is that mixing them naively sounds terrible. Two full-volume audio sources fight each other: the music muddies the source audio, the source audio muddies the music, and the result is worse than either alone.

What the pros do: ducking

The fix is ducking — the same technique radio DJs and podcasters use. When there's something worth hearing in the source audio, the music drops back. When there isn't, the music comes back up to full.

Done by hand in a normal editor, that means drawing volume keyframes across the whole timeline: down at the start of each moment, back up at the end, smoothed so it doesn't click. On a four-minute edit with dozens of moments, that's genuinely tedious work — which is why most people either skip the source audio entirely or just leave it fighting the music.

Doing it automatically

Onset Engine does the whole envelope for you. Tick Mix Original Audio in the project settings before you render, and it will:

1. Pull the source audio from every clip in the edit and stitch it to the timeline
2. Scan it for where the audio is actually active — using an adaptive threshold, so it adjusts to how loud your particular source material is rather than a fixed cutoff
3. Duck the music to 30% across those stretches, with 150ms crossfades either side so transitions glide instead of clicking
4. Ignore anything under about a third of a second, so brief noise doesn't cause the music to stutter up and down

You'll see it report what it found in the console — something like [DUCK] 47 active segments detected for ducking. That number is a useful sanity check on its own (more below).

Getting a good result

Pick source clips worth hearing

Ducking makes the source audio audible — which is only an improvement if it's worth being audible. Clips with clean, present audio work. Clips with a TV in the background, a noisy room, or a camera-mic hum become a lot more noticeable once the music steps back for them.

Match the track to the technique

Dense, wall-of-sound tracks fight this: the music has to duck so far to make room that the drop is jarring. Sparser tracks with space in them — slower tempo, less mid-range clutter — leave room for the source audio to sit in, and the ducking becomes almost invisible.

Read the segment count

The console number tells you a lot before you even listen:

• Very high (hundreds on a short edit) — it's likely triggering on background noise, so the music will feel restless. Use cleaner source clips.
• Very low (a handful) — barely any of your source audio is registering as active, so you're near enough to a music-only edit. Check your clips actually have usable audio.

Think about where it lands

Ducking is at its best in the quieter stretches — the build, the verses, the sections between drops. On a full-energy drop you usually want the music at full force anyway. If your driver puts the busiest source audio exactly where the track peaks, the two are competing for the same moment; shifting that content to the lower tiers usually sounds better.

Worth knowing

• It's a render-time setting, so you can render both ways from the same edit and compare.
• The 30% duck level is fixed in the automatic pass. If you want a specific mix, render with source audio and finish in a normal editor.
• Ducking follows the source audio, not the beat. It reacts to what's in your clips, so it won't dip in time with the music — that's the intent.
• Both audio sources carry the same copyright considerations as ever — see the copyright guide.