Clipping, Limiting
& Loudness

What mastering is — and what it isn't

Mastering is about translation: your track sounding right everywhere, from car stereo to AirPods to club PA to laptop speaker. Right tonal balance, genre-appropriate loudness, enough dynamic range to survive streaming normalization intact. It's not about making your track as loud as possible. That war ended with CDs.

Every major streaming platform normalizes integrated loudness. Too loud and it gets turned down; below target and it gets left alone (platforms generally don't boost quiet tracks). Hitting the target or coming in slightly under is usually the right call. Going louder mostly just costs dynamic range.

LUFS, True Peak, and the numbers that matter

LUFS measures perceived loudness with frequency weighting, which is closer to how your ears work than raw RMS. Two tracks at identical RMS can sound wildly different in loudness. LUFS is the only loudness number that matters for delivery.

True Peak vs sample peak

Your DAW meter shows sample peak. But when the file gets decoded, the signal reconstructed between samples can go higher. Those inter-sample peaks clip DAC outputs and get worse through MP3/AAC codecs. Set your limiter ceiling to −1 dBTP, not −1 dBFS. Different measurements.

Interactive platform reference

Set your track's integrated LUFS below and see how each platform handles it. Toggle to True Peak for the ceiling reference.

Platform targets: the exact numbers

The mastering chain: signal flow

1. Linear Phase EQ    — corrective cuts, subtle air shelf (+1–2 dB @ 16 kHz)
2. Multiband Comp     — tighten 100–300 Hz mud, 2:1 ratio, slow attack
3. Saturation/Clipper — add harmonics, reduce peak-to-RMS ratio, 0.5–2 dB of gain
4. Stereo Imager      — widen slightly (M/S), ensure sub is mono below 80 Hz
5. Limiter            — true peak ceiling −1 dBTP, target −9 to −7 LUFS for EDM

Clipping vs limiting: the fundamental difference

Hard digital clipping

Hard clipping truncates the waveform at the maximum value: flat top, squared-off peaks, heavy odd-order harmonic distortion (3rd, 5th, 7th). On a kick drum, 0.5–1 dB of this adds punch. On a pad or synth lead, it sounds broken.

Intentional clippers like Kazrog True Iron, iZotope Ozone Clipper, and DMG Limitless (clip mode) let you control exactly how much flat-topping happens, with oversampling to keep aliasing under control.

Soft / analog-style clipping

Soft clipping applies an S-curve transfer function that progressively rounds off peaks instead of chopping them, producing even-order harmonic distortion (2nd and 4th), the same thing tubes and tape do. The Sonnox Oxford Inflator, Slate FG-X, and iZotope Ozone Maximizer in soft mode all work here.

Limiters

A brickwall limiter is a compressor with infinite ratio and a very fast attack (0.05–1 ms). It reads ahead via a lookahead window and starts applying gain reduction before the peak arrives. Waveform shape stays intact; only amplitude gets reduced. The tradeoff: the gain recovery after each transient can audibly pump if you drive it hard. A clipper has no release time: it works sample-by-sample.

Crest factor: the number that links it all

Crest Factor (dB) = Peak (dBFS) − RMS (dBFS)

Example:
  Peak  = −0.3 dBFS
  RMS   = −14.2 dBFS
  CF    = −0.3 − (−14.2) = 13.9 dB   ← typical unmastered electronic mix

After a clipper stage:
  Peak  = −1.0 dBFS
  RMS   = −8.6 dBFS
  CF    = −1.0 − (−8.6) = 7.6 dB    ← mastered EDM track

A typical unmastered electronic mix sits at 12–15 dB CF. A well-mastered EDM track is 6–9 dB. The clipper shaves peaks; average level rises to fill the headroom.

The limiter deep dive

Lookahead

Limiters read 0.5–5 ms ahead and start applying gain reduction before the peak hits. Without it, even a 1 ms attack lets some of the transient through. Keep lookahead off live monitoring channels (it adds latency), but always on for mastering exports. Above 2 ms can smear transients.

Attack and release

Start with adaptive modes (Pro-L2's "Aggressive" or "Modern"). Going manual: 0.05–0.3 ms attack catches transients; 30–100 ms release keeps pumping off sustained material. High-end limiters like Pro-L2, Weiss DS1-MK3, and DMG Limitless also offer transient preservation, which lets brief kick attacks behave differently from sustained loud sections.

Oversampling

4x minimum. A 4x oversampled limiter can detect inter-sample peaks that fall between samples in the original signal. 8x gives marginal extra benefit at significant CPU cost.

Concrete settings for electronic music

FabFilter Pro-L2 (Electronic/EDM starting point):
  Ceiling:          −1.0 dBTP      ← true peak, not sample peak
  Output Level:    −1.0 dBTP       ← match ceiling for proper gain staging
  Lookahead:        0.5–1 ms       ← 1 ms for safety, 0.5 ms for less smear
  Transient:        20–40%         ← 40% preserves punch; 20% = more loudness
  Attack:           Adaptive       ← or 0.1 ms manual for tighter transients
  Release:          Adaptive       ← or 30–80 ms manual for electronic material
  Oversampling:     4x minimum     ← 8x for critical listening passes
  Algorithm:        Modern         ← transparent; Aggressive for more color

Aim for gain reduction meter: 1–3 dB typical, 5 dB absolute maximum.
If hitting more than 5 dB GR, pull back input gain or add clipper before.

Transparent vs character limiters

Loudness targets by genre

Target the loudness that fits how and where your track gets played. An ambient track mastered at −8 LUFS will get turned down 6 dB on Spotify, which tends to expose a quiet noise floor. A techno record at −16 LUFS can feel thin next to everything else in a DJ set. Neither is a disaster, but matching genre norms usually makes sense.

Common mastering mistakes

1. 1
   Over-limiting: the pumping, squashed dynamic. Pushing a limiter past 4–6 dB of gain reduction creates audible pumping. Fix it: reduce input level, add a clipper upstream, or accept a lower LUFS.
2. 2
   Ignoring true peak. Mastering to a 0 dBFS sample peak ceiling without true peak mode guarantees inter-sample clips that crack during AAC or MP3 encoding. Set the ceiling to −1 dBTP, enable true peak detection, and verify with a true peak meter (iZotope Insight 2, Nugen MasterCheck, Youlean Loudness Meter Pro) before delivery.
3. 3
   Over-processing in the midrange. Aggressive multiband compression in the 200–800 Hz range creates a flat, tonally dead midrange. On a decent mix, the multiband should barely touch the mids. If you're hitting 3–4 dB of GR on a midrange band, that's a mix problem.
4. 4
   Ignoring mono compatibility. Electronic music gets played on mono systems constantly: phones, Bluetooth speakers, club PAs with center-summed subs. An over-widened stereo image can phase-cancel badly in mono, pulling out low-mid weight or the fundamental of a bass sound. Check mono before you print. Every time.
5. 5
   Not referencing on multiple playback systems. At minimum: studio monitors, open-back headphones (AKG K702, Sennheiser HD600), earbuds, a laptop speaker, and a car. The laptop check specifically tells you whether your low-end translates or just becomes a muddy rumble.
6. 6
   Skipping dithering on the final bounce. Rendering to 16-bit? Dithering goes last, after the limiter, with nothing after it. Without it, the quantization error from the bit depth reduction is correlated and audible in quiet passages as a harsh, grainy quality. Noise shaping (Waves L2's UV22HR, iZotope's MBIT+) pushes the dither noise above 15 kHz.

Mastering cheat sheet

1.  Linear Phase EQ    → corrective cuts, air shelf (+1–2 dB @ 16 kHz)
2.  Multiband Comp     → 100–300 Hz mud, 2:1 ratio, ≤ 2 dB GR per band
3.  Saturation/Clipper → crest factor reduction, 0.5–2 dB of drive
4.  Stereo Imager      → M/S width, mono sub below 80 Hz
5.  Limiter            → −1.0 dBTP ceiling, target integrated LUFS for genre
6.  Dithering          → TPDF + noise shaping on 16-bit renders only