- Kiloeve Reviews
- May 8, 2021
- 7 min read
Updated: May 16, 2021
Continuation From the Previous Post
In the previous post, I discussed the fundamentals of how compression works in decreasing the dynamics in audio tracks. I will now continue on with how this is applied to limiters and clippers.
Limiting
If you want to control the sound of a signal without changing its envelope, you use a limiter. A limiter is a really hard compressor. Anything that is 12:1 or above is considered limiting, although others may say it is as low as 10:1 or as high as 20:1. The attack time on a limiter is also zero milliseconds or non-existent, as it is much faster than a compressor.
Unfortunately, though limiters are simpler than compressors, they are all much more different from each other and the parameters aren’t as easy to define. The definitions are much more interchangeable and less prescriptive.
THRESHOLD - Same as a compressor. However, some “adaptive limiters” will have automatic gain compensation (i.e Loudmax and W1 Limiter). This means when the threshold of a limiter is decreased, the output is increased as well, whilst it is working really hard to avoid the audio from distorting. A “good” limiter does this transparently, so that you really have to try and listen out for the audio being crushed.
GAIN - Some compressors may label the GAIN as Input, such as VladG Limiter No6. Input on a limiter is identical to a compressor. Again, it’s the same as boosting on a gain or trim plugin, or in other words, the amount of audio you are feeding in. In an “adaptive” limiter, it will adapt to what’s happening in the signal to make sure that the quiet parts meant to remain quiet aren’t raised as heavily, but the loud parts are raised more. This all depends on how much you push it.
RELEASE - Same definition of the release in a compressor.
LOOKAHEAD - You may see this on a certain limiter that ‘judges’ the entire signal and looks ahead for any ‘nasty’ surprises. In other words, it looks ahead to see when it needs to quickly cut down on the peaks. A limiter that has a lookahead is more likely to work harder on louder sections and do less work on quieter sections. If you set it long enough, you are likely to achieve smoother limiting. However, if you set it too long, you will have very noticeable pumping effects, which may lead to additional artefacts.
THRESHOLD, INPUT GAIN AND OUTPUT GAIN (separated) - Some limiters may have a threshold, an input knob AND an output knob, such as TDR’s Limiter 6. This may be good for gain staging and getting an honest result, especially in mixing, to see just how hard you are limiting those transients. However, in mastering, sometimes you just want the limiting aspect to just make the music louder. This is exactly why I covered and compared the W1 Limiter and LoudMax in previous posts to TDR’s Limiter 6 and VladG’s Limiter No6, because of the very few parameters they have.
In mastering especially, a limiter is one tool that helps achieve commercially loud song, therefore level matching is not necessarily as important as it is with compression. This is exactly why I covered the W1 Limiter and LoudMax in previous posts, as they only have two parameters to control.
CEILING - Think of the ceiling as the ceiling of a room. It stops MOST (99.9%) of the signal from going above your set ceiling. You can set your ceiling anywhere, but it is bad practice to set it at exactly 0dBFS full scale or above. However, a common misconception is that you have to set your true peak at -2dBFS. The reason you don’t have to do this is because most converters will introduce intersample peaks anyway, (peaks you don’t see in your DAW) sometimes a few decibels above what was intended. Also, regardless if people rip MP3s or WAV files from the internet, or buy them from a digital platform like Spotify, it is likely that most professional engineers ignore the guidelines and let their true peak go into the red. And even if a mastering engineer avoided going over 0dBFS in the mastering session, it is likely that an aggregator has somehow introduced distortion, that all of us are initially unaware of. However, if you set your ceiling in between -0.1dBFS to -0.5dBFS, (and sometimes lower if needed) you will get less of these intersample peaks. However, avoid setting the ceiling too low. Certain loud genres such as metal and drum and bass, for example, will sound noticeably different due to some sonic qualities in the music that are affected from lowering the ceiling. For a post on different loudness meters, please see the “Is Your Loudness Meter Lying to You? Part 1 and 2” for more details, especially regarding true peak.
Clipping
Clipping is arguably the most interesting out of the three processes. A common myth spread around is that a clipper is the same as a compressor. This statement is FALSE, especially in the digital domain. A clipper has no attack or release time. Typically, a whole sound can become almost ‘flatlined’ in extreme cases. In terms of the difference between a limiter and a clipper, a limiter still has a very tiny bit of audio that goes above the ceiling, but a clipper is a kind of limiter that shaves off the transients, so that NOTHING goes over it.
Clipping introduces headroom, whilst not changing much of the original sound. If you are subtle enough, it will sound almost identical, IF you compensate it with GAIN.
However, if you use too much of it, you will start to notice artifacts, and these are elements of distortion. Distortion occurs because of the artificial harmonics you are introducing into the signal. This may be a good or a bad thing, depending on what sounds you are applying the clipper into. On transient sounds such as drums and percussion, this may bring out more of the punch. Other instruments such as a piano, violin and acoustic guitar, for example, could possibly make the distortion too noticeable and distasteful. This all depends on what you aim to achieve on your music.
There are typically only two parameters on a clipper:
GAIN (sometimes called ‘saturation’) - How much distortion you are adding. On many clippers, you can possibly get away with up to 9 or 10dB of gain to drums and synths, for example, before the results sound “bad” or like they don’t belong. This depends on the sound source itself and in this case, the type of drum you are dealing with.
OUTPUT - The output that you pull down after adding clipping. It should typically be equivalent to the amount of gain you have put in, if you are purely aiming for headroom and subtle distortion. If you want an obvious effect, then the output is left.
Additional parameters you may see on other clippers are:
OVERSAMPLING - A useful process for minimising aliasing, depending on how high you set this parameter. The higher the oversampling, the smoother the sound, but the more CPU you end up using. Some clippers let you change the amount of oversampling (1x, 2x, 4x, 8x, etc). In VladG’s Limiter No6, you can use the “SIGNAL” feature in the clipping section to let the oversampling process come first, before the signal gets clipped. This is one of the only clippers that does this system properly and promotes a cleaner sound. More information about the SIGNAL function in VladG’s Limiter No6 in this video at 17 minutes:25 seconds.
SHAPE - Some clippers let you change how soft or how hard the harmonics are. On one end, you have soft clipping, which introduces fewer harmonics. On the other end, you have hard clipping, which can be hard to pull off digitally if your gain staging is not up to standard, as it is not a typically pleasing sound. Even in an analogue realm, achieving a good dose of hard clipping can be challenging.
CEILING - Same concept as the ceiling on a limiter, but lets absolutely nothing go through. A limiter will still let a very small amount of audio through, if you are running the signal hot.
Many skeptics say that clipping is bad because it introduces distortion, and while this is true, it is not the conventional distortion that you think of. Actual distortion occurs over 0dB full scale, but if you think of something like a square wave, distortion is already occurring in the actual waveform. It’s just not going over zero.
However, like with compression and limiting, you should be careful to avoid overusing it. Unlike compression and limiting, it is advised that you avoid using clipping on a mix bus, or a summed track. This is because of a phenomenon called intermodulation distortion, and you are more likely to get this if you feed multiple signals to another channel.
In the hyperlink I have provided above, Nate from Reframinator illustrates intermodulation distortion. In the video, there are three sine tones on three separate tracks, all different in their hertz: 100Hz, 125Hz, 150Hz. He applied a clipper on each track, and the harmonics sounded adequately in place. If the pitch is to be played around with, the harmonics also follow along accordingly. There is no intermodulation distortion, as of yet.
However, he illustrated the exact same sine tones whilst summing up these tracks (or in other words, applying a clipper to the mix bus instead). This time, you will notice that there are nasty harmonics all over the place, with all three tones turned on at the same time. Unless your music has just intonation or pure intonation, or you are aiming for that sound, it will most likely be difficult for you to get away with this result.
Here is one SOLOED sine tone, in a summed track:
TWO sine tones playing at once, in a summed track:
ALL THREE sine tones playing at once, in a summed track:
Summary
Compressor - If you’re hoping to contain a sound and control peaks, use a compressor. A compressor will typically compress, or turn down dynamics. It has an attack and release RATE.
Limiter - If you want to control a sound’s peaks without changing its envelope, you most likely need a limiter. It has ZERO attack, and it typically has a release.
Clipper - If you want to shave off the peaks, use a clipper. It will typically add harmonics, and if not applied heavily, the results can sound transparent. It does not automatically work on every instrument. A normal clipper has no attack and no release.
Thank you for reading this two-part post on dynamics processing. If you managed to grasp just one concept from this post and the hyperlinks, then you already know a lot about dynamics processing. As you now know, compressors, limiters and clippers are a significant portion of dynamics processing and take a big chunk in mixing as a whole. There are two other main processors, noise gates, expanders, and upwards compression typically used for increasing dynamics, which I will cover in the future.
-Alex (KiloeveReviews)
