How to Clean Up a Vocal Recording: The Right Order of Operations

Clean a vocal in this order: hum and noise, plosives, clicks and mouth noise, breaths, sibilance, then tuning and timing. The order matters more than the specific plugins, because every corrective process changes what the next one hears. Fix a plosive after you have compressed, and you have already amplified the problem into the rest of the chain.

What you need

• The raw vocal take, unprocessed
• A DAW with EQ, a de-esser, and clip gain or volume automation
• Optionally a repair suite such as iZotope RX, or the equivalent tools in your DAW
• Headphones for detail work, speakers for judgment

Step 1: Fix what you can at the source first

Before any processing: if you can re-record, do. Thirty minutes of moving the mic, adding a blanket behind the singer, and getting the pop filter positioned properly beats hours of repair.

Specifically — most home vocal problems are one of: too far from the mic, a reflective wall directly in front of the singer, a mic pointed at a computer fan, or gain set too low so the noise floor comes up with the level.

Step 2: Remove hum and broadband noise

Hum sits at 50 or 60Hz plus its harmonics, depending on your mains frequency. A dedicated hum removal tool or a set of narrow notch filters at the fundamental and each harmonic will take it out cleanly.

Broadband noise — fans, air conditioning, preamp hiss — needs a spectral noise reduction tool. Capture a noise profile from a pause in the take, then apply the smallest amount that works. Over-applied noise reduction produces a metallic, watery quality on the vocal itself that is worse than the noise was.

Do this first, because everything downstream will amplify whatever noise you leave.

Step 3: High-pass, carefully

Roll off below about 80Hz for most voices, 100Hz for a lighter voice. You are removing rumble, handling noise, and floor vibration, none of which carry vocal information.

Use a gentle slope and check on a system with real low end. Cutting too high thins the chest of the voice.

Step 4: Fix plosives individually

A plosive is a burst of low-frequency energy from a P or B. The best fix is per-instance, not a global filter:

• Find the plosive on the waveform — it looks like a large low-frequency lump
• Draw clip gain down over just that syllable, or apply a steep high-pass to that region only

A global high-pass steep enough to kill plosives will thin the whole vocal. Handle them one at a time.

Step 5: Remove clicks and mouth noise

Zoom in and either use a de-click tool or manually redraw the waveform. Mouth clicks are short and localised, so manual repair is genuinely fast once you can spot them.

Hydration and a break fix this faster than any plugin, for future takes.

Step 6: Manage breaths

Select each breath and reduce it by 6 to 12dB with clip gain, or use a breath control tool. Do not delete them.

A vocal with no breaths sounds like it was assembled rather than sung. Listeners cannot always say what is wrong, but they hear it.

Step 7: De-ess

Now, not earlier — because noise reduction and EQ both change the sibilant range.

Set the de-esser's detection band around 5 to 9kHz for most voices, and aim for 2 to 4dB of reduction on the worst syllables. If it is working constantly, the threshold is too low.

Split-band de-essing sounds more natural than wideband on most material. And check the result soloed and in context, because a de-esser that sounds right soloed is often too aggressive in the mix.

Step 8: Address room reverb, with realistic expectations

De-reverb tools estimate the reflections and subtract them. On a slightly live room they work well. On a genuinely reverberant space they leave a hollow, phasey artifact that is more distracting than the room was.

Try a small amount. If it needs a lot, the answer is re-recording, not more processing.

Step 9: Only now, tuning and timing

Pitch correction after cleanup, not before — tuning algorithms track cleaner audio more accurately. Melodyne, Auto-Tune, or your DAW's built-in tool all work; the choice is about workflow, not quality.

Same for timing. Align consonants, not vowel onsets, when tightening doubles against a lead.

The order, as a checklist

Everything after this is mixing, not cleanup.

Related reading: recording vocals at home, mixing vocals guide, and home studio acoustics basics.