How to Clean Up Video Audio (Isolate, EQ & Loudness)
Viewers forgive shaky footage; they won’t forgive bad audio. Here’s how to clean up your video’s sound on your phone — voice isolation, EQ, dynamics, and the loudness step almost everyone skips.
Audio is half of video — and the half people notice when it’s wrong. Muddy voices, music that drowns the talking, and clips that are too quiet all read as “amateur” instantly. The fixes are quick, and they live right on the clip.
The three-step audio cleanup
- Isolate a clean voice out of noise.
- Shape it with EQ and dynamics.
- Even out the loudness for the platform.
Do them in that order: isolate first so every later move lands on a clean signal, not on noise.
Step 1 — Isolate a clean voice
If the dialogue was recorded in a noisy place, the fastest noise-reduction move on a phone is stem separation. Select the clip, tap Stems, and ExpoCut splits the audio into separate tracks — Vocals, Drums, Bass and Other — entirely on-device. Mute everything except the voice and the background hiss, rumble and music drop away.
Vocals · Drums · Bass — each on its own track
Audio FX
For the full breakdown of stem separation — including the Studio vs Fast quality modes — see stem separation & vocal isolation.
Step 2 — Shape it with EQ and dynamics
With the voice on its own track, open Audio FX. The fastest win is the Clean Voice preset — one tap optimizes a voice for clear narration. From there, EQ & Tone lets you roll off low rumble and add a touch of presence so words cut through, while Spatial and Modulation add polish.
Clean Voice — one-tap cleanup
EQ & Tone · Dynamics · Spatial · Modulation
Balance the music under the voice. Once the voice is clean, drop the music track’s volume in the talking sections to roughly 30% so it sits underneath, then bring it back up in the gaps. Your voice should always be clearly on top.
Step 3 — Even out the loudness
Here’s the step almost everyone skips. Platforms play everything at a target loudness; if your video is quieter than the feed it feels weak, and if it’s louder it gets turned down (and distorts). In the Dynamics tab, a Compressor evens out a voice that jumps between loud and quiet — “consistent loudness” — and a Limiter stops peaks from exceeding a safe ceiling.
Dynamics
Compressor & Limiter
Where to go next
- Pull voice out of noise: stem separation & vocal isolation.
- Build the music mix: add music & audio.
- Ship it at the right level: best export settings.
Great audio is invisible — viewers only notice when it’s missing. Isolate, shape, even out, and your video will sound as finished as it looks.
Frequently asked questions
Can I reduce background noise on my phone?
Yes — isolate the voice using stem separation to pull speech away from background noise, then shape what's left with EQ and dynamics. It won't beat a treated studio, but it rescues a lot of real-world audio, all on-device.
How do I make music quieter when someone is talking?
Lower the music's volume in the section where the voice plays — pull it down to roughly 30% so it sits under the speech, then bring it back up in the gaps. Isolating the voice onto its own track first makes this balance much easier to hear.
Why does my video sound quieter than others on Instagram?
Because it isn't evened out to a consistent loudness. Run a compressor and limiter on the voice so it sits at a steady level before export — no surprises when it autoplays next to everything else in the feed.
Sound as good as you look
Voice isolation, EQ, dynamics and loudness — clean, consistent audio that survives the feed. Fix the half of video everyone hears.
Get ExpoCut Browse all guides