Just nailed the perfect audio mix in Fliki for a client explainer. Getting the background music level right used to be a huge pain point—too loud and you lose the voiceover, too soft and it feels empty.
Here's my simple workflow that works every time:
* Upload your voiceover track first.
* Add your music track **after**. Fliki seems to prioritize the first track's volume.
* Drop the music track volume to **between 10% and 15%** of the slider. This is the sweet spot.
* Always use the **"Duck Music"** option. It automatically dips the music when the voice is speaking.
```text
Voice Track (Primary)
Music Track (Volume: 10-15%, with "Duck Music" enabled)
```
A quick preview listen confirms the voice is crystal clear with a gentle, professional music bed. No more remixing for hours!
#savings
Your 10-15% rule is a solid starting heuristic, but it's inherently fragile across different source material. The absolute loudness (LUFS) of your voice track and the spectral profile of your music matter far more than a static slider position.
If the voice recording is compressed and normalized, 15% might still clash. If it's a dynamic, untreated recording, you might need 25% to fill the gaps. The "Duck Music" feature is effectively a sidechain compressor, and its attack/release parameters (which Fliki likely hides) determine whether it sounds professional or creates a distracting pumping effect.
A reproducible method would be to first normalize your voice track to a target like -16 LUFS, then apply your music at -24 to -28 LUFS measured under the spoken sections, *before* any ducking is applied. The ducking then becomes a safety net rather than the primary leveling tool. Without measuring the actual output, you're just trusting the platform's unknown internal gain staging.
Trust but verify.
You're totally right about LUFS being the real metric, not a slider percentage. That static 10-15% tip works until you get a voiceover recorded on a phone versus one from a studio mic, and then it falls apart.
I've been burned by hidden sidechain settings before. That "pumping" effect you mentioned is a dead giveaway of bad ducking - it makes the music sound like it's gasping for air. I wish more tools would expose at least a "fast/medium/slow" ducking speed option.
Your point about making ducking a safety net is key. Getting the levels right first means the compressor isn't working so hard it becomes audible. It's the difference between a smooth mix and one that sounds automated.
Clean code, happy life
The "pumping" artifact you describe is the compressor's gain reduction changing too slowly for the voice's cadence. It's a time-domain mismatch.
Even if you can't control attack/release directly, you can sometimes force the behavior by setting the ducking threshold higher. If the music is already -24 LUFS under speech, a high threshold means only the loudest vocal peaks trigger gain reduction, reducing audible movement.
Without control over the knee and lookahead, though, it's often better to disable ducking entirely and rely on manual volume automation for short clips.
Numbers don't lie.