Why Loudness Matters More Than You Think

You have probably had the experience of settling into a podcast or video, only to lunge for the volume dial because the presenter is suddenly deafening. Or the opposite: you crank everything up, still strain to hear, and then an advert blasts you out of your seat. These moments are rarely about the quality of the microphone or the skill of the edit. They are about audio levels.

Consistent loudness is one of the quiet markers of professional work. Get it right and your audience simply relaxes into what you have made. Get it wrong and they are distracted, fatigued, or reaching for the skip button. The good news is that a handful of straightforward habits will keep your projects sounding polished, whether you are producing a weekly podcast, a client case study video, or a series of social clips.

The Numbers That Matter: Peaks and Headroom

Digital audio is measured in decibels relative to full scale, written as dBFS. Zero dBFS is the absolute ceiling; go above it and the signal is clipped, which sounds harsh and crackly. You never want to be anywhere near that ceiling while recording or mixing.

A practical target for most speech-led content is to keep your peaks around minus six decibels. That leaves plenty of headroom for unexpected moments: a laugh, a sudden emphasis, a chair creaking. If your loudest peaks sit at minus six, quieter passages still have room to breathe without disappearing into the noise floor.

• Peaks: aim for roughly minus six dBFS during recording and mixing
• Average level: conversational speech usually sits comfortably between minus eighteen and minus twelve dBFS
• Noise floor: keep room tone and hiss below about minus fifty dBFS so it does not become audible when you raise the level
• True peak ceiling: leave at least one decibel of space below zero on your final export

Those figures are not rules carved in stone, but they are a reliable starting point for spoken-word audio. Music beds and stingers can sit lower still, so they support the voice rather than compete with it.

Clipping, Distortion and the Damage You Cannot Undo

Clipping happens when a signal exceeds the maximum your system can represent. The waveform gets flattened, and the result is a brittle, buzzing edge that is particularly unpleasant on the human voice. Crucially, clipped audio is effectively destroyed. You cannot restore the shape of a waveform that was never captured, so any repair is really just cosmetic. Prevention is everything.

Watch for two forms of clipping. The first is obvious: meters hitting red in your recording software or on your interface. The second is subtler, happening inside a cheap preamp or a camera's built-in input, where the meter looks fine but the sound is already brittle. If in doubt, record a test, listen back on headphones, and trust your ears over the display.

Setting Levels Before You Press Record

Good levels begin long before the mix. Spend two minutes on setup and you will save an hour of frustration later.

• Set gain on the interface, not in software. Adjust the dial so normal speech peaks around minus twelve to minus six dBFS
• Ask your guest to speak at their loudest. A laugh or an excited aside is the true test of your headroom
• Use a limiter cautiously. A gentle one can catch stray spikes, but it is not a substitute for sensible gain staging
• Record at 24-bit. The extra dynamic range gives you far more flexibility when you come to tidy things up
• Monitor with headphones. Speakers in an untreated room will flatter or hide problems you need to hear

If you are recording multiple people, set each microphone individually. One loud guest and one soft-spoken guest need different treatment at the source, not just in the edit.

Mixing, Normalising and Final Delivery

Once your edit is tidy, it is time to balance and finish. Ride the levels so that every voice feels present and consistent, cutting the odd loud syllable rather than squashing the whole track. Gentle compression can help even out speech, but reach for it after you have fixed the obvious peaks manually.

When the mix sounds right, normalise the finished file. Normalising raises or lowers the entire piece so it hits a target loudness, preserving the internal balance you have crafted. For podcasts and spoken-word video, a useful reference is around minus sixteen LUFS integrated, with true peaks no higher than minus one decibel. Many platforms will turn your audio down if it arrives hotter than their preference, so there is no advantage in pushing it.

• Listen on more than one system: headphones, laptop speakers, a phone
• Check the whole piece, not just the opening
• Export consistently: same loudness target for every episode or video in a series

A Quick Checklist Before You Publish

Before you upload, run through a short routine. Are peaks sitting around minus six decibels? Is there any clipping you can see or hear? Does the loudness stay even from start to finish? Have you normalised the final export and left a decibel of headroom at the top?

Do this consistently and something quietly valuable happens: your audience stops noticing the sound and starts paying attention to what you are actually saying. That is the whole point of good audio levels, and it is well within reach for any small business or independent creator.