Audio Repair Guides - Remove Noise, Hum, Clicks & Clipping

To clean up a bad recording, work from the loudest, most obvious problem down to the subtle ones, and fix them in the right order: dehum and denoise first to clear the background, then declick to remove pops and crackle, then declip to rebuild distorted peaks, and only then reach for tone shaping like de-muffling or de-essing. Repair is subtraction, not creation. You can pull a voice out from under noise, tame a hum, and smooth out clicks, but you cannot invent detail that the microphone never captured. The goal is not perfection, it is a recording that sounds intentional instead of broken. Run one clean pass, listen on real speakers, and stop before you start chewing up the parts you actually want to keep.

Every recording carries a mix of problems, and the biggest mistake people make is attacking all of them at once with one aggressive setting. This guide walks through each type of damage, what actually causes it, how much of it is realistically recoverable, and the order that gives you the cleanest result. Where a topic deserves a deeper walkthrough, we link out to a focused guide. If you just want to hear the difference right now, drop your file into Sauce Audio Repair and let it handle noise, hum, clicks and clipping in a single pass.

What counts as a "bad" recording, and can it actually be fixed?

A bad recording is rarely bad for one reason. It is usually two or three faults stacked on top of each other: a noisy room, a cheap interface humming at the wall frequency, a mouth click or two, and a singer who pushed the input into the red. Each of these is a separate problem with a separate fix. Lumping them together is why so many "cleaned up" files end up sounding underwater.

Here is the honest split. Broadband noise, hum, clicks, and mild clipping are the repairable end of the spectrum. Tools have gotten genuinely good at these, and the results can be close to invisible when you go easy. Heavy reverb, badly muffled audio, and severe distortion are the hard end. You can improve them, sometimes dramatically, but you cannot fully undo them, and pretending otherwise leads to artifacts that are worse than the original flaw. Set your expectations before you touch a single control and you will make better decisions.

How do you remove background and broadband noise?

Background noise is the constant, hissy, wide-frequency bed under a recording: air conditioning, computer fans, street rumble, mic self-noise, the general hum of a room that is not treated. Because it is broadband, it sits across the whole spectrum, which is exactly why a simple EQ cut cannot remove it without gutting your voice too.

Denoise works by learning what the noise looks like and subtracting it, moment to moment. Modern denoise is spectral and often model-based, so it can tell the difference between a steady noise floor and the signal you want. The trap is greed. Push denoise too hard and you get that watery, gurgling, "underwater" artifact, plus a voice that has lost its air and life. Aim to knock the noise down until it stops distracting you, not until it is gone entirely. A little residual floor sounds natural; a vacuum-sealed voice sounds processed. For the full walkthrough, see how to remove background noise from audio.

How do you get rid of hum and buzz?

Hum is different from broadband noise, and it needs a different tool. Electrical hum sits at a specific frequency tied to your mains power: 50 Hz in most of the world, 60 Hz in North America, plus a stack of harmonics above it (100/120, 150/180, and so on). It comes from ground loops, unbalanced cables, dimmer switches, cheap power supplies, and gear placed too close to the recording chain.

Because hum lives at known, narrow frequencies, a dehum tool can surgically notch out the fundamental and its harmonics while leaving the rest of the audio almost untouched. This is one of the most satisfying repairs, because a good dehum is nearly invisible and the buzz just disappears. The key is matching the right base frequency to where the recording was made: guess 60 Hz on a 50 Hz recording and you will leave the real hum sitting there. Our dedicated guide to how to remove hum from audio covers spotting harmonics and choosing the base frequency.

How do you remove clicks, pops, and crackle?

Clicks are short, sharp transient faults: mouth clicks and lip smacks on a vocal, pops from digital dropouts, crackle from a bad connection or an old vinyl transfer, and the plosive thumps that hit a mic when someone says a hard "P" up close. Unlike noise and hum, these are not constant. They are isolated events, and that is what makes them fixable with high precision.

A declick tool scans for these transients and repairs each one by interpolating across the damaged sample, essentially redrawing the waveform where the click was so smoothly that your ear never notices the patch. Declick is safe to use fairly assertively on genuine clicks, but be careful: set it too aggressive and it can start softening real transients like consonants, snare hits, or pick attacks, which dulls the whole recording. Match the tool to the fault. Deep dive here: how to remove clicks and pops from audio.

What can you do about clipping and distortion?

Clipping happens when a signal is recorded louder than the system can represent, so the peaks get sliced flat. Those flattened tops are missing information, and that missing information is heard as harsh, crunchy distortion. This is the repair with the hardest limits, so it is worth being precise about what is and is not recoverable.

Mild, occasional clipping is genuinely fixable. A declip tool estimates the shape the waveform would have had and reconstructs the missing peak, then you pull the level down to give it headroom. The result can be remarkably clean because the clip only lasted a fraction of a second and there was plenty of intact audio around it to model from. Heavy, sustained clipping is a different story. When large stretches are flattened, there is simply too little real signal left to rebuild, and no tool can invent it convincingly. You can reduce the harshness and make it more listenable, but the distortion character will linger. The lesson is brutal and simple: clipping is far easier to prevent than to repair. For the specifics, read how to fix clipped audio, and for distortion that is not strictly clipping, how to fix distorted audio.

Can you remove reverb and roominess?

This is the question people most want a "yes" to, and it is the one where honesty matters most. Reverb is not a layer sitting on top of your voice. It is your voice, reflected off walls, floor, and ceiling, arriving at the mic hundreds of times at slightly different delays and volumes, all tangled together with the direct sound. There is no clean line to cut along, which is why removing reverb is fundamentally harder than removing noise or hum.

De-reverb tools have improved a lot, and they can meaningfully reduce roominess, tighten up a boxy recording, and pull a voice forward. But push them and you hear the cost: a pinched, phasey, slightly robotic tone as the tool guesses wrong about which energy is direct and which is reflected. Treat de-reverb as damage reduction, not removal. Get it 70 percent of the way there and stop. The real fix is upstream: record closer to the mic, in a smaller or more treated space, so the reflections never dominate in the first place. More detail in how to remove reverb from a recording.

How do you fix muffled or dull audio?

Muffled audio sounds like it is coming through a blanket: no top-end sparkle, no clarity, everything soft and closed-in. The cause is missing high frequencies, whether from a mic pointed the wrong way, a pop filter or clothing over the capsule, a phone in a pocket, or a heavy low-pass somewhere in the chain.

The important thing to understand is whether the highs were captured at all. If there is real high-frequency content that is just too quiet, a high-shelf EQ boost or a de-muffle tool can bring it back to life and the recording opens right up. If those frequencies were never recorded, boosting only lifts hiss, because there is nothing musical up there to amplify. Some modern tools can synthesize plausible high-end to fill the gap, and they can help, but they are guessing. Start with a gentle shelf, listen for clarity versus harshness, and stop when it sounds present rather than brittle. Walkthrough here: how to fix muffled audio.

How do you tame harsh sibilance?

Sibilance is the sharp, whistling "sss" and "shh" energy that spikes on certain consonants, usually sitting somewhere between 5 and 10 kHz. A little is natural and keeps speech intelligible. Too much is fatiguing and stabby, and it gets worse after you brighten a dull recording or lean on compression, both of which push those peaks higher.

The right tool is a de-esser, which reacts only when the sibilant energy spikes and leaves the rest of the vocal alone, rather than a static EQ cut that dulls every word. Set the frequency where the harshness lives, then use the least reduction that removes the sting. Over-de-essing gives you a lisp, so back off the moment the "s" sounds start to disappear. Sibilance control is a core part of finishing a voice, which is why it lives inside broader vocal work like how to clean up a vocal recording.

How do you rescue a phone recording or a room recording?

Phone and handheld room recordings are the ultimate stress test, because they usually combine every problem at once: broadband noise, room reverb, muffled highs from a covered mic, aggressive built-in compression, and sometimes clipping when someone spoke too loudly. There is no single button for this. You have to work through the layers in order.

Start by knocking down the noise floor and any hum, then declick anything sharp, then gently de-reverb to tighten the space, and finish with careful tone shaping to recover intelligibility and control sibilance. Go easy at every stage, because the faults compound. A phone recording will never sound like a studio, but a methodical pass can move it from "unusable" to "clearly understandable," which is usually the actual goal. See how to make a phone recording sound better for the full sequence, and if it is specifically a voice you are trying to save, how to fix a bad vocal recording.

What is the right order to repair audio in?

Order matters more than any single setting, because each fix changes what the next tool sees. Denoise before dehum and you make the tonal hum harder to isolate. Boost the highs before you denoise and you amplify the hiss you are about to fight. Here is a reliable sequence:

• Dehum first. Notch out the 50/60 Hz fundamental and harmonics while the signal is still untouched.
• Denoise second. Pull the broadband floor down now that the tonal hum is gone.
• Declick third. Remove pops, mouth clicks, and crackle from the cleaned signal.
• Declip fourth. Rebuild distorted peaks, then drop the level for headroom.
• De-reverb fifth. Tighten roominess once the background is clean.
• Tone and de-ess last. Fix muffle, add presence, and tame sibilance on a clean, controlled signal.

Repair problems first, tone and polish second. If you shape the sound before the noise is gone, you spend the whole session fighting yourself. Sauce Audio Repair handles the first four stages in one pass so you can jump straight to the finishing work, and Vocal Polish takes it from clean to release-ready with presence, warmth and level.

What can and cannot be fixed, realistically?

Keep a clear map of what is worth your time. You can reliably reduce broadband noise, remove hum and buzz, repair clicks and pops, rebuild mild clipping, and add clarity when the highs actually exist in the file. You can partially improve heavy reverb, severe distortion, and audio that was muffled at the source, but you cannot fully restore them, and chasing perfection there just trades one flaw for a worse-sounding one.

What you can never do is recover information that was never captured: detail lost to a covered mic, a voice buried under a room that swallowed it, or peaks obliterated by hard clipping. Repair moves things toward acceptable. It does not turn a phone memo into a studio session. Knowing where that ceiling sits keeps you from over-processing, which is the single most common way people make a rough recording sound worse.

How do you avoid needing repair at all?

Every fix in this guide is a compromise, and the cheapest compromise is the one you never have to make. Prevention at the source beats any tool. Set your input level so the loudest moment peaks around -6 dB, leaving headroom so nothing clips. Record as close to the mic as the source allows to drown out the room. Kill obvious noise before you hit record: fans, chargers, chatty computers. Use balanced cables and a decent power supply to keep hum out. Add a pop filter and a little distance to control plosives and sibilance.

Do those five things and most recordings need only the lightest touch, or none at all. When you do need to clean something up, come in with realistic expectations, work in the right order, and use the gentlest setting that solves the problem. That mindset, more than any single plugin, is what separates a recording that sounds repaired from one that just sounds good.