Why dual camera video sounds distorted or crackly in loud places

What is actually happening: the microphone runs out of headroom

A microphone and the circuitry behind it can only represent sound up to a certain level. Under that ceiling, a recording gets louder as the source gets louder, faithfully. Above it, the top and bottom of the waveform stop following the sound and get flattened instead — the peaks are cut off rather than continuing to rise. That flattening is what you hear as a harsh, fuzzy, crackling sound, and it is a distinctly different problem from audio that is simply loud: loud-but-clean audio still traces the shape of the sound; clipped audio has had that shape cut off.

This happens during the analog-to-digital conversion, which is a property of the hardware, not of any recording app running on top of it. By the time a video file exists, the clipping already happened — there was never a moment where an app could have caught it and reacted.

Why there is nothing to turn down beforehand

A dual-camera app records straight from whichever audio input iOS has selected as the default — the built-in microphone, or an external one if one is connected — with no manual gain or volume dial sitting in front of it. That absence is deliberate rather than an oversight: a manual level is easy to leave in the wrong position between takes, and the phone’s own automatic level handling already manages normal speaking distances reasonably well without asking you to think about it.

The catch is that automatic level handling reacts to sound over time — it is built to settle at a sensible level across a scene, not to anticipate a single instant. A burst of applause, a bass hit, a firework or someone shouting right next to the phone can clip before any automatic system has a chance to respond, precisely because it is a sudden peak rather than a sustained loud passage. There being no gain dial to leave in the wrong place also means there is none you could have preset low enough to guarantee safety.

What actually helps while you are shooting

Since there is no setting to lower, the levers that work are physical, and they are usually about one specific source rather than the whole scene. Loud environments rarely clip evenly — a crowd recorded from a normal distance is generally fine, and the moment that distorts is almost always one identifiable thing: a speaker stack, someone shouting directly at the phone, an amp, a burst of fireworks close by.

• Put distance between the phone and that one loud source, even a small amount. Sound pressure falls off quickly with distance, so a few extra feet from a speaker or an amp is often the difference between clean and clipped.
• Angle the phone away from a point source instead of pointing the microphone straight at it — a PA speaker or someone shouting directly into the mic is a much harder hit than the same sound arriving from an angle.
• If you are using an external microphone, it becomes the system’s default input the same way the built-in one does — the app has no separate handling for it. A lavalier clipped to your collar for your own voice is usually fine; a microphone aimed straight at a PA speaker or an amp carries the same clipping risk as the phone’s own mic, just relocated.
• If the loudest moment is predictable — a chorus, a countdown, a firework show — start the take with the phone already at a safer distance rather than moving in and hoping it stays clean.

Why boosting it afterward cannot bring it back

Quiet audio and clipped audio look similar as problems but are not similar to fix. Quiet audio is small but intact — the actual shape of the sound is still there in the file, just at a low level, so raising it evenly in an editor recovers something close to what happened in the room. Clipping is different: once the top of the waveform is flattened, the information about how loud that instant actually was is gone. It was never captured in the first place, not just captured small.

Raising the volume of a clipped section in an editor afterward makes the crackle louder along with everything else — it does not remove it, because there is no clean version underneath to reveal. Some editing tools include a de-clip or restoration effect that smooths the flattened peaks into something less harsh to listen to, and that can help a section feel less abrasive, but it is a cosmetic reduction of the artifact, not a recovery of the original sound. Getting the distance right during the take is the only thing that actually prevents this, which is worth knowing before a loud event rather than after one.