| Recent Comments |
| Categories |
| Archives |
| Tags |
What is the difference between a limiter, maximizer, and a clipper?
Clipping (Figure 1) occurs when the audio signal amplitude exceeds the maximum voltage (or quantizing) capability of the system. As a result, the output signal stops abruptly at the voltage (or quantizing) limit, and so the tops and bottoms of the audio wave form are, in effect, ‘sawn off’. This is something that inevitably produces strong harmonic distortion. The system is entirely linear up to the clipping level, and then horribly non-linear, so signals slightly lower than the clipping level are passed unaffected, while those at, or higher than, the clipping level are mangled, resulting in the addition of complex additional harmonics that were not present in the original.
In digital systems (e.g., DAWs), we can often recognize clipping it because it produces unnatural anharmonic distortion through aliasing. This is especially obvious for signals with a well-defined harmonic structure, such as pianos, voices, and so on. Noise-like sources, such as cymbals and snare drums, when aliased, produce more noise-like distortions, so although the timbre might change slightly it won’t usually be recognized as distortion.
With limiting a loud signal is briefly attenuated, specifically to avoid clipping. The short-term dynamic changes involved do introduce a form of distortion, but it is a far more benign form and, again, it is rarely recognized as such. The system is linear up to the limiting level, above which the waveform is reduced in amplitude but more or less retains its original shape, and thus remains harmonically intact. The distortion that results is usually negligible. Limiters are basically high ratio compressors with fast attack and release times. Generally, some limiters give you control over attack, release, look-ahead, and other parameters. See SoundOnSound for a discussion of release and limiting. Hard (brickwall) clipping is limiting with zero attack and release time and is the same as clipping.
Figure 1. Although clipping and limiting are similar processes, they are not the same thing. As can be seen from the picture, clipping abruptly stops the signal from going over a maximum voltage limit, causing distortion. Limiting is a far more controlled process in which the signal is attenuated specifically to avoid clipping (copied from SoundOnSound, 2013)
Limiting and clipping both introduce distortion products. Soft clipping (Figure 2) has a less aggressive onset of clipping: essentially a combination of limiting prior to clipping. Soft clipping is a type of distortion that saturates an audio signal and rounds off the peaks that pass 0db. This gives it a warm, analog sound, and is different to hard clipping, which sounds more digital – cutting off the peaks that pass 0db off in a straight line, square wave fashion.
Figure 2. Hard clipping vs soft clipping (copied from WippedCreamSounds).
An audio maximizer is like a limiter but specifically designed to bring a full mix to an optimum loudness level and control its peak digital level. While a limiter simply knocks down or chops off the loudest peaks, a maximizer increases the loudness of a track and at the same time sets a ceiling for its peak level to prevent clipping. A limiter’s job is to set a ceiling while the maximizer pushes the music up to the ceiling. Professional models offer look-ahead processing and true-peak (inter-sample peak) limiting. Many maximizer plugins also provide oversampling options to avoid digital aliasing distortion. Every maximizer uses a limiting algorithm as part of its process, but many limiters do not have the full features of a maximizer. Maximizers also have controls not found on typical limiters, like transient enhancement, true-peak protection, oversampling, and dither.
Clipping and Brickwall limiting
(Reference: Clipping in Mastering…..Let’s Talk; GearSpace)
Clipping and brickwall limiting are both stopping peaks from exceeding a defined threshold. While limiting does this with very fast attack and release times (and lookahead) to try to reduce distortion, clipping is instantaneous and produces more distortion artefacts. Limiting can sound smoother and firmer, but can also dull transients. Clipping can keep more of the energy and attack in transients, but can get hard, harsh, and aggressive sounding. Generally, clipping produces odd-order harmonics, which most people hear as the opposite of pleasing or smoothing, although it can sound more exciting.
When pushing loudness, clipping and limiting can work well in tandem. How much you use of each will be different for each mix. Say you’re reducing peaks by a total of 4dB. Some mixes will sound better with 3dB of clipping and 1dB of limiting, while others will sound better with this ratio reversed. Then other mixes will sound better with about 2dB of both, or full clipping or full limiting.
It’s a balancing act between the aesthetic qualities of each process (described above) and how they interact with the mix. For some mixes, limiting will be more transparent or musical, while for others clipping will be more transparent or musical.
Tokyo Dawn Records (the make of TDR Limiter 6 GE) takes a different perspective:
The audibly and technically optimal way to control the overshoot of tonal content is the traditional limiter. The optimal way to control transients in noisy spectra is the clipper.
Music usually is a mix of both, but by varying degrees. If the overshoots you’re trying to control are mostly transient and/or noise alike, use a clipper. If you have to control tonal content, e.g. sustained tones, use a limiter.
Issues appear once clippers face tonal content (strongly distorting of the original timbre), or limiters have to control noisy/transient content (causing strong long term intermodulation distortion, e.g. dulling, breathing or pumping, also between L and R).
In proper mixes, overshoots tend to purely consist of noisy transients, permitting the material to clip gracefully. Such mixes can be mastered with a bit of clipping only. This conservative structure also helps with translation to classic radio or TV broadcast.
But as soon you intent to dig deeper into the tonal region (usually more or less equivalent with the RMS), more complex control schemes/automatisms become necessary, some hybrid between limiting and clipping.
Some types of signals can be clipped almost without loss. Take dynamic percussions for example, one can often clip them with double-digit GR and still sound ok. But tonal content of course better doesn’t get clipped.
Modern limiters seamlessly transition between clipping and limiting behavior in dependence of the incoming signal.
Some clipping and limiting at the mix bus is also done: first clip to shave some transients to provide a bit more headroom before limiting.
References
The text above was copied and adapted from the following sources:
