Definition
The crest factor is the ratio between the peak value and the effective (RMS) value of a signal:
C = |x|_max / x_rms, in dB C_dB = 20 · log10(|x|_max / x_rms) = L_peak − L_rms.
The equivalent term in electronics is PAPR (peak-to-average power ratio), expressed in dB with the same value. Reference values for ideal waveforms:
| Signal | Crest factor |
|---|---|
| Square wave | 1 (0 dB) |
| Sine wave | √2 (3.01 dB) |
| Triangle wave, sawtooth | √3 (4.77 dB) |
| Gaussian noise (white or pink) | 10–14 dB over musical durations |
The crest factor depends on the measurement window: calculated over the entire file it includes the variations between sections and the silences; calculated over short blocks (300 ms–3 s) it describes the local microdynamics. With the AES17 RMS convention (full-scale sine wave = 0 dBFS RMS) the crest factor comes out 3.01 dB lower than that calculated with the mathematical convention.
Parameters and units
- Unit: dB (or dimensionless ratio).
- Window: entire file, 3 s blocks, 300 ms windows.
- Typical values over the entire file (mathematical convention):
| Material | Crest factor |
|---|---|
| Uncompressed drum track | 18–24 dB |
| Recorded vocal | 12–18 dB |
| DI electric bass | 10–14 dB |
| Sine-wave sub and 808 | 4–8 dB |
| Pre-mastering mix | 12–16 dB |
| Modern pop master | 7–10 dB |
| Very loud EDM master | 5–8 dB |
Musical function
The crest factor indicates how much the transients stand out above the average level: it is a direct index of impact and of how much a signal has been compressed or limited. It is used to choose the amount of compression and limiting, to compare tracks and masters with references and to predict how much margin a signal requires in gain staging. A low crest factor does not necessarily mean loud: a sine-wave sub track has a crest factor of 4–6 dB by nature.
Practical example
- Kick: peak −6 dBFS, RMS −24 dBFS → crest factor 18 dB. After a 4:1 compressor with a 1 ms attack the kick has RMS −22 dBFS and peak −10 dBFS: crest factor 12 dB, less prominent transient. With a 20 ms attack the peak stays at −6 dBFS and the RMS rises with the make-up gain: crest factor 16 dB, transient preserved.
- Mix and master: pre-mastering mix with peak −3.2 dBFS and RMS −15.8 dBFS (crest 12.6 dB); master with peak −1.0 dBFS and RMS −9.2 dBFS (crest 8.2 dB). Mastering removed 4.4 dB of crest factor.
- Silence: the same master with 30 s of silence added to 3 minutes has an RMS 0.7 dB lower and a crest factor 0.7 dB higher, with no musical difference whatsoever.
Common mistakes
- Comparing values calculated with different RMS conventions (a 3 dB discrepancy).
- Measuring files with long silences or reverb tails and reading an inflated crest factor.
- Comparing the crest factor of songs with different spectra: a mix with a lot of sub naturally has a lower crest factor.
- Chasing a low crest factor to gain loudness: perceived loudness also depends on the spectrum and the density of the arrangement.
- Treating crest factor and PLR as equivalent: the former uses sample peak and unweighted RMS, PLR uses true peak and K-weighted loudness.
How to check it
- App analysis: the Crest factor (peak − RMS) row is the difference between Sample peak and RMS calculated over the entire file and on both channels, with the mathematical RMS convention. The PLR (true peak − integrated LUFS) provides the same information on a perceptual basis.
- Before and after: analyze the track in the app before and after a compressor or a transient shaper to measure how much crest factor has been removed or added.
- Meters: Wave Candy (FL Studio) and meters with simultaneous peak and RMS readout show the difference in real time.
Starting settings
| Parameter | Value | Why |
|---|---|---|
| Pre-mastering mix | 12–16 dB | Intact transients before limiting. |
| Pop master | 7–10 dB | Commercial density with residual impact. |
| Reduction in mastering | 3–6 dB | Beyond this, the transients flatten audibly. |
| Uncompressed drums | 18–24 dB | Reference for evaluating the compression applied. |
| RMS convention | stated (mathematical or AES17) | Avoids 3 dB discrepancies in comparisons. |
| Measurement file | without leading and trailing silences | Silences raise the value for no musical reason. |
How to check the result
- Crest factor (peak − RMS) of the pre-mastering mix ≥ 12 dB in pop and rock.
- Crest factor reduction between mix and master ≤ 6 dB.
- Crest factor of the master within ±1.5 dB of the reference in the same genre, measured without silences.
- The crest factor is read together with PLR and LRA, not on its own.