Dynamic range compression is the process of reducing the difference between the loudest and quietest parts of an audio signal by automatically lowering gain when the level exceeds a threshold. In mastering, it is applied to a finished stereo mix to glue the elements together, control peaks, increase perceived density and prepare the track for its loudness target, while preserving as much of the music’s natural movement as possible.
The term describes the process, not the device. It is distinct from data compression such as MP3 or AAC encoding, which reduces file size and does not intentionally change the level relationship between loud and quiet passages.
How dynamic range compression changes level
Compression works through a detector and a gain control element. The detector measures the incoming level, and when it rises above the threshold, the gain element reduces it according to the ratio. At a ratio of 2:1, a signal that exceeds the threshold by 6 dB comes out only 3 dB above it.
Because loud parts are lowered, the overall level can then be raised with makeup gain. The quiet passages become relatively louder, so the effective dynamic range shrinks. Time constants decide how quickly this happens: attack sets how fast gain reduction begins, and release sets how fast it recovers.
Detection type also shapes the sound. Peak detection reacts to instantaneous levels and controls transients, while RMS detection responds to average energy and behaves more like human loudness perception. The knee determines whether compression starts abruptly at the threshold or gradually around it.
The following table lists the main parameters with ranges commonly used in mastering, which are much gentler than typical mixing settings.
| Parameter | Typical mastering range | Effect on the master |
|---|---|---|
| Ratio | 1.2:1 to 2:1 | Low ratios keep the music natural |
| Gain reduction | About 1 to 3 dB | More reduction flattens dynamics audibly |
| Attack | 10 to 50 ms | Slower attack lets transients through |
| Release | 100 ms to auto | Should follow the tempo and groove |
| Knee | Soft, 6 to 12 dB | Smooth onset hides the processing |
| Sidechain high-pass | 60 to 150 Hz | Stops bass from driving the whole mix |
Compression approaches used in mastering
Mastering engineers choose between several compression architectures, each solving a different problem in a finished mix. The choice depends on whether the issue is overall consistency, a problem in one frequency band or a lack of density.
| Approach | How it works | Best suited to |
|---|---|---|
| Broadband bus compression | One gain change for the whole signal | Glue and gentle overall control |
| Multiband compression | Separate compressors per frequency band | A boomy low end or harsh upper mids |
| Parallel compression | Blends a heavily compressed copy with the dry mix | Density while keeping transients |
| Mid-side compression | Processes center and sides separately | Controlling a dominant vocal or wide elements |
| Limiting | Very high ratio with fast attack | Final peak control before delivery |
In a typical chain, gentle broadband compression comes before equalization or after it depending on the problem, and a true peak limiter comes last. Splitting the work across stages, each removing only a little, usually sounds more transparent than one processor working hard.
Compression, limiting and loudness normalization
Dynamic range compression, limiting and loudness normalization are often confused, but they act on different aspects of a master. Compression reshapes dynamics over time, limiting controls only the highest peaks, and normalization changes the overall playback level without altering dynamics.
| Process | Changes dynamics | Main measurement | Purpose |
|---|---|---|---|
| Compression | Yes, moderately | Gain reduction in dB | Density and consistency |
| Limiting | Yes, at peaks only | True peak in dBTP | Ceiling protection and loudness |
| Loudness normalization | No | Integrated LUFS | Consistent playback level |
| Clipping | Yes, by distortion | Sample peak | Aggressive peak shaving |
Streaming platforms and broadcasters apply loudness normalization, commonly around -14 to -16 LUFS for music services and -23 LUFS under EBU R128 for European broadcast. A master that has been compressed and limited to -7 LUFS will simply be turned down. It then loses its loudness advantage while keeping the reduced punch, which is the central argument against excessive compression in modern mastering.
Measuring how much dynamic range remains
Engineers judge compression by meters as well as by ear. Useful indicators include the peak-to-loudness ratio (PLR), which compares true peak to integrated loudness, and loudness range (LRA), which describes how much the loudness varies across a track in LU.
A PLR of roughly 8 to 12 dB generally indicates a master that still has clear transients, whereas values near 5 dB suggest heavy compression and limiting. These figures vary by genre: an orchestral recording may need a wide LRA, while electronic dance music is often intentionally dense. No single number defines a correct master; the meters only confirm what the engineer hears.
Pitfalls of over-compression
Excessive compression is the most common mastering fault because louder initially sounds better in a side-by-side comparison. Level-matched listening is therefore essential before judging any setting.
- Pumping and breathing occur when the release fights the tempo, making the background rise and fall audibly.
- Dulled transients come from attack times that are too fast, which flattens drums and removes punch.
- Bass-driven gain changes happen when the kick triggers compression of the whole mix; a sidechain high-pass filter prevents this.
- Listening fatigue results from a constantly dense signal with no contrast between sections.
- Inter-sample overs can appear after heavy processing; a final true peak limiter at about -1 dBTP protects lossy encodes.
In simple terms
Dynamic range compression narrows the gap between loud and soft sounds. Imagine someone watching a film with a remote control, turning the volume down during explosions and up during whispers. A compressor does the same thing automatically and very quickly.
In mastering, it is used gently so a song sounds full and even on any speaker without losing its energy. Used too much, it makes music sound flat and tiring.
A practical example
A mastering engineer receives an acoustic pop song whose choruses feel noticeably weaker than the verses on small speakers. The mix itself is balanced, but the dynamics swing widely and the final chorus lacks weight.
The engineer applies broadband compression at a 1.5:1 ratio with a 30 ms attack, an auto release and a 100 Hz sidechain high-pass filter, aiming for only 1.5 to 2 dB of gain reduction on the loudest passages. A true peak limiter follows with a ceiling of -1 dBTP.
The finished master measures about -12 LUFS integrated with a PLR near 10 dB. The choruses now sit consistently forward, the acoustic guitar transients remain crisp, and the song keeps its impact after streaming normalization.