Skip to content

Dynamic range control: how to manage levels, peaks and punch

Send feedback

Last updated 8 min readTags:MixingMasteringDynamics

Learn how to control audio dynamics with automation, compression and limiting, from recording levels to the final master.

Dynamic range control means adjusting how much an audio signal's level changes over time. It includes evening out a vocal, shaping a drum hit and controlling the difference between a quiet verse and a loud chorus.

The right treatment depends on the problem. A phrase that disappears may need automation. A sharp peak may need compression or limiting. A chorus that feels too restrained may need less processing.

Start with what you hear, then use meters to check the result.

What does dynamic range tell you?

In music, dynamics can describe changes within a single sound or across an entire arrangement. The initial hit of a snare and the rise from verse to chorus happen on different timescales. One measurement cannot describe both fully.

Equipment specifications use dynamic range differently: the usable span between the noise floor and the maximum level the system can handle. This matters when recording, but it does not tell you how lively a performance sounds.

These measurements answer different questions:

Measurement What it tells you Useful for
Peak level, in dBFS The highest sample level Checking proximity to digital full scale
True peak, in dBTP An estimate of the highest level of the reconstructed waveform, including between samples Checking peaks that a sample meter may miss
Integrated loudness, in LUFS Average loudness across the measured programme, with perceptual weighting and gating Comparing whole-track loudness
Loudness Range, or LRA, in LU The spread of short-term loudness values after gating Assessing longer-term loudness variation
Crest factor, in dB The difference between peak and RMS level over the same measurement interval Assessing peaks relative to average signal level

LRA is a statistical measure, not simply the loudest moment minus the quietest. It excludes very quiet material and extremes. It also says little about the attack of individual drum hits. The EBU's Loudness Range specification explains the calculation.

Choose a tool for the problem

Compression is one form of dynamic range control. It is not the answer to every level problem.

Tool What it changes Typical use
Clip gain or volume automation The level of selected words, notes or sections Raising a quiet vocal phrase or lowering an unusually loud note
Compressor Gain in response to the detected signal level Controlling repeated level changes or shaping attack and sustain
Limiter Peak level, with strong gain reduction near a ceiling Catching peaks or controlling the final output
Downward expander or gate Level below a threshold Reducing noise or microphone bleed between notes and hits
Transient shaper The attack and sustain of a sound Making a drum hit sharper or shortening its tail
Clipper The shape and height of waveform peaks Reducing brief peaks where the resulting distortion is acceptable
Dynamic EQ or multiband compressor Level within selected frequency ranges Controlling intermittent boom or harshness

A gate may reduce the signal to silence or by a limited amount, depending on its range setting. Neither a gate nor an expander can separate wanted sound from bleed while both are present at similar levels.

A transient shaper can change peak and average levels, so check its output. A clipper reduces peaks by reshaping them; it does not preserve crest factor. Its distortion may be unobtrusive on some material and obvious on other material.

1. Record with room for louder moments

Set recording gain while the performer plays or sings the loudest expected passage. Leave enough space below clipping for an unexpectedly strong note.

With a modern 24-bit recording system, there is usually no reason to record close to 0 dBFS. Focusrite's recording guidance makes the same point: lower recording levels provide room for louder notes.

There is no universal −18 dBFS RMS recording target. A sustained synth and a snare drum can have very different peak levels at the same RMS reading. Judge available headroom from the peaks, not the average alone.

If the recording is noisy, check the room, microphone position, source level and preamp gain. Turning up a recorded file raises the captured signal and its noise together. It does not improve that recording's signal-to-noise ratio.

Clipping at the input cannot be undone simply by turning the file down. Repair software may estimate damaged peaks, but a clean recording is preferable.

2. Fix uneven phrases before compressing

Listen to the track in the mix. Identify whether the problem affects a whole phrase, individual notes or only brief peaks.

For example, a vocal may have a quiet verse and a much louder chorus. One compressor setting might barely affect the verse but heavily compress the chorus. Use clip gain before the compressor to reduce that difference, then set compression for the remaining variation.

Use volume automation later in the chain when you want to change the final balance without changing how hard the vocal hits the compressor. Check your DAW's signal routing; clip gain and fader automation do not necessarily act at the same point.

Keep expressive changes that suit the performance. The aim is for the listener to hear the phrase clearly, rather than for every word to reach the same level.

3. Set compression by listening to attack and recovery

A conventional downward compressor reduces gain when the detected level exceeds its threshold. Threshold and ratio determine how strongly it responds. Attack and release determine how that response develops over time.

Listen for these changes:

  • Attack: Does the initial hit remain clear, or has it become too soft? A faster attack can control sharp peaks; a slower attack may retain more of the onset.
  • Release: Does the sound recover naturally, or does its level swell or dip between notes? Check whether gain reduction continues into the next hit.
  • Threshold and ratio: Are uneven notes more consistent? If the performance has become lifeless, try less gain reduction.

These controls can change the balance between attack and sustain, not just reduce overall variation. FabFilter's compressor guide demonstrates those effects.

Match the processed and bypassed versions to roughly the same perceived loudness. Extra output gain can make an unhelpful setting seem better.

Bus and parallel compression

Bus compression applies a shared gain change to a group of tracks. On drums, that can make the kit feel more connected. It can also make the cymbals dip whenever the kick hits. Listen for both effects before deciding whether to keep it.

Parallel compression blends an unprocessed signal with a compressed version. It can add sustain while retaining some of the original attack. The combined signal's crest factor may still fall as its average level rises. Check the blend at matched loudness.

4. Investigate what is driving the mix-bus processing

If the whole mix dips whenever the kick or bass plays, inspect the low end before increasing compression or limiting. A few strong notes may be driving gain reduction across the entire mix.

Try lowering those notes with automation or controlling them on the source track. If only a frequency range becomes excessive, dynamic EQ or multiband compression may be appropriate. A constant tonal imbalance may need an ordinary EQ adjustment instead.

For a mix going to mastering, avoid accidental clipping and leave room for further processing. An exact −6 dBFS peak level is not compulsory. A clean mix can be turned down before mastering without changing its internal balance, as explained in iZotope's headroom guide.

Follow the mastering engineer's delivery instructions. Keep processing that is integral to the sound, and discuss a separate version without final loudness limiting if useful.

5. Judge the master at matched loudness

Mastering may include compression, EQ, clipping and limiting. Each process should earn its place. Heavy limiting can soften attacks, distort bass notes or make the whole track audibly rise and fall.

Compare the master with the mix at similar perceived loudness. Listen to a dense chorus, an exposed passage and a section with strong bass. Does the processed version still have the impact and contrast you want?

Crest factor can help describe the change, but it is not a quality score. A high reading may reflect one stray peak. A lower reading may suit a dense arrangement. Use consistent meter settings when comparing versions, and listen to the affected sounds.

What streaming normalisation changes

Loudness normalisation adjusts playback gain. It does not undo compression or restore flattened attacks. Two tracks with similar integrated loudness can still sound different because of their tonal balance, arrangement and dynamics.

Spotify currently uses −14 LUFS for its Normal setting on supported playback. Its published guidance recommends a −14 LUFS master with true peaks below −1 dBTP, or below −2 dBTP for masters louder than −14 LUFS. These are recommendations rather than a universal delivery specification for music streaming. Playback settings and devices vary.

Choose the dynamics and loudness that suit the track, then check the relevant delivery guidance. A separate master for each service is not automatically necessary.

A final listening check

Before exporting, check that:

  • Quiet phrases remain audible without losing their intended restraint.
  • Strong notes do not cause unwanted level dips elsewhere in the mix.
  • Drum attacks have the weight and definition you want.
  • Processing has not introduced distracting pumping, distortion or noise.
  • The chorus still provides the intended contrast with the verse.
  • Peak levels and file format meet the agreed delivery requirements.

If processing sounds better only when it is louder, repeat the comparison with the levels matched.

Was this page helpful?

    Type to search titles, headings and summaries.