How to Fix a Muddy Mix

Production guides · Meme Plugins · 26 October 2026

To fix a muddy mix without making it thin, identify the source of the buildup, improve the balance, check the reverbs, then make the smallest useful EQ change. Preserve the weight that belongs to the song. Mud can come from overlapping low and low-mid energy, but also from sounds that sustain too long. iZotope’s guide to muddy mixes explains both causes. The workflow below applies those principles to a typical mixing session.

1. Find the moment, then find the part

Loop a passage where the mix loses definition. Listen for something specific: bass notes merging, a vocal losing its words, or chords becoming a continuous cloud. Mute one supporting part at a time while the rest plays. If removing a pad restores clarity, bring it back lower before adding EQ. Try a simpler voicing, a different register or fewer overlapping notes. iZotope’s unmasking guidance starts with thoughtful arrangement: several similar parts sharing a range can create avoidable competition.

2. Give kick and bass clear roles

Listen to kick and bass together, then add the rest of the mix. Decide which should provide the deepest weight and which should carry more audible attack or note definition. That choice depends on the sounds and arrangement; neither instrument automatically owns a fixed frequency. Try changing the kick sample, shortening an overlong note or adjusting a fader. If each part sounds impressive alone but they blur together, solve their relationship before boosting both. The same masking principles apply.

3. Check the space around the sound

Mute reverb returns briefly. If clarity returns, restore them one at a time and lower the send or return level. A separate wet return lets you shape the reverberation without removing body from the dry instrument. Filter unnecessary lows from the reverb path, or make a gentle cut where its tail crowds the mix. Shorten decay if notes keep hanging over the next phrase. iZotope’s reverb guide describes separate reverb EQ and timing controls. Listen through the gaps: the fog may be lingering after the instrument has stopped.

4. Choose static or dynamic EQ deliberately

For a steady tonal buildup, try a modest bell cut on the contributing track while the full mix plays. A narrow cut suits a specific resonance; a broader problem may need a broader, shallower adjustment. Stop if the instrument loses its useful body.

For a problem that appears only on certain notes, consider dynamic EQ. Its gain changes with signal level, so attenuation can occur when the problem becomes prominent instead of remaining constant. Set the threshold and maximum reduction conservatively, then hear the whole phrase. FabFilter’s dynamic EQ documentation explains the level-dependent behavior. Judge the resulting sound rather than the amount of movement on the display.

5. Compare at a similar loudness

Bypass the change and adjust output gain so the processed and original versions have similar perceived loudness. Listen for clearer separation while the vocal, snare and bass retain their substance. FabFilter documents output gain for compensating EQ level changes, and warns that its automatic gain compensation is an estimate rather than a measured loudness match. Check by ear as well. For a wider perspective, follow our reference-track comparison guide and use The Reckoning alongside listening.

6. Use Cleansing as a controlled experiment

Keep your original file. In The Cleansing, start with “1 Is it muddy?” For processing, choose “2 Clean a track” or “3 Clean a vocal,” then try CLEANSE before EXTREME CLEANSE. Use A/B to judge the result before downloading the cleaned WAV. Compare that export with the original at a similar loudness. Keep the version that serves the song; a more dramatic change earns no extra blessing.

Frequently asked questions

What frequency should I cut to remove mud?

There is no universal cut frequency. Find the track and passage causing the problem, then listen for the range where a reduction improves definition while preserving useful warmth.

Why did my mix become thin after removing mud?

The cuts may have removed body from several instruments at once. Reduce or bypass them, reassess the balance and reverb, then restore only the adjustments that help the full mix.

If clearing congestion leaves a separate sharp or tiring sound, use the harsh-mix workflow to find its source and timing. Treat that edge as a separate problem so a low-mid cleanup does not lead to an unnecessary cut across all useful brightness.

Before removing more low-frequency material, follow the high-pass filtering workflow. Compare the same instrument in the busy section and a quieter one so cleanup preserves the body the song still needs.


Browse all production guides → · Free browser audio tools

An original mix through ordinary CLEANSE

We uploaded an original, synthesized four-bar mix to Cleansing and used the ordinary CLEANSE action, without the extreme option. The source contains drums, bass, keys, melody and overlapping lower pad voices. These players compare that source with the actual downloaded result.

Before: original synthetic mix, 100 BPM, 9.6 seconds.

Download the original source WAV (1.69 MB).

After: actual CLEANSE result, with RMS matched to the source for comparison.

Download the RMS-matched comparison WAV (1.84 MB).

The untouched export is 2.081 dB lower in whole-file RMS than the source. The comparison copy applies only a constant +2.081 dB gain to that export; the source is unchanged. RMS matching aligns average signal level, rather than guaranteeing identical perceived loudness. The matched copy peaks at −6.519 dBFS, with no full-scale PCM samples.

Play or download the untouched Cleansing export
Actual downloaded output, with no later gain adjustment.

Download the untouched CLEANSE result WAV (1.84 MB).

What the measurement shows: in this example, absolute energy from 200 to 500 Hz fell 27.50% (−1.397 dB) in the untouched export. Once RMS was matched, energy in that band was 17.07% higher (+0.684 dB) than the source. This example therefore does not demonstrate a selective reduction of the 200–500 Hz band. Listen to the matched pair to decide how the change suits this material.

The lower 80–200 Hz band did fall after RMS matching: 8.45% (−0.384 dB). The 500–1000 Hz band rose 32.74% (+1.230 dB). This is a transparent listening example of a changed tonal balance, rather than proof that every mix will sound better.

Saved-file measurements; all versions are stereo, 16-bit PCM and 9.6 seconds long.
VersionSample rateSample peakWhole-file RMS
Original source44.1 kHz−6.000 dBFS−19.542 dBFS
Untouched CLEANSE export48 kHz−8.600 dBFS−21.623 dBFS
RMS-matched comparison48 kHz−6.519 dBFS−19.542 dBFS
How these figures were measured

Measurement method: each native-rate channel was demeaned, windowed with a symmetric Hann over the full recording, and zero-padded for a 524,288-point FFT. One-sided power was normalized by FFT size and window energy; bin centers within each stated band were summed, including the lower edge and excluding the upper edge, then the two channel powers were averaged. The band figures are window-weighted estimates. Peak and RMS come directly from saved PCM, not true-peak or LUFS measurements. This is one synthetic example; no Reckoning result is included.