Limiter

A “lookahead” dynamic range limiter to stop peaks above the set threshold value.

Unlike the Soft Clipping Limiter, Broadcast Limiter and LADSPA limiter, this limiter does not use wave shaping distortion to limit the peaks. It is a very fast compressor and is able to limit the maximum peaks without introducing high levels of harmonic distortion.

The limiter is very simple to use and requires little explanation, but for anyone that is unfamiliar with limiters, here’s a few tips.


Limit to (dB): This control sets the maximum peak level. As peaks in the original audio approach this level the gain is reduced so as to prevent the peaks exceeding the set level.

Hold (ms): Holds the gain at the reduced level after a peak is detected so as to prevent the gain from “riding the waveform” which would cause harmonic distortion.
Shorter Hold times allow the peaks to be tracked more accurately and the limiter will respond faster to the dynamics.
If there are high levels of very low bass it will be necessary to increase the Hold time to avoid distortion. The default 10 ms hold time is sufficient for frequencies down to 100 Hz without distortion. To cleanly limit high amplitude, very low frequency bass (down to 50 Hz) the Hold should be increased to 20 ms. Setting the hold to 50 ms is sufficient right down to 20 Hz but the delay before the gain level “recovers” is likely to be too slow for most material.

Make-up Gain (0=No, 1=Yes): When enabled (default) the output is amplified by an amount equal to the “Limit” level. If the input audio has a peak level of 0 dB, the peak output level will also be 0 dB. When disabled the peaks are limited only.

Additional notes:

Stereo Tracks: As is normal for this type of effect, the left/right channels of a stereo track are processed independently.

Lookahead: This limiter looks ahead for peaks and will begin to change the gain just before the peak occurs. This ensures that all peaks, no matter how fast they occur, will be caught. The lookahead time is roughly a quarter of the hold time.

Knee: This limiter uses a relatively “soft knee” so as to avoid unnecessary distortion. That is, the amount of compression (compression ratio) progressively increases as the input gets louder. At the “Limit to” level the compression ratio is infinite (brick wall) which ensures that peaks will not exceed the limit.

Creative use: This limiter can be used on its own, or can be used to limit peaks after running a compressor that does not use lookahead (such as the SC4 LADSPA compressor). This can produce “crisper” compression than using a lookahead compressor such as the standard Audacity compressor or Chris’s dynamics compressor.

Over 0 dB input: The input waveform should not exceed 0 dB. Over 0 dB input signals are illegal and will be hard clipped to 0 dB before processing with the limiter. If necessary, the Amplify or Normalize effects should be run before applying this limiter to ensure that the input does not exceed 0 dB.

Current download on Wiki Download Nyquist Plug-ins page:
http://wiki.audacityteam.org/wiki/Nyquist_Effect_Plug-ins#Limiter

There is also a “type 3” version that has a drop-down menu to select Make-up gain (for Audacity 1.3.x or later)
Limiter (2) on Wiki Download Nyquist Plug-ins page:
http://wiki.audacityteam.org/wiki/Nyquist_Effect_Plug-ins#Limiter_.282.29

OK, I’ve tried this with a few sources and it’s simply AWESOME. It’s great to maximize loudness of your tracks without the ugly distortion you get with digital clipping or hard limiting, and also without the extra harmonics that a soft-clipping limiter creates. It manages to avoid both because, just like Steve said, it works as an extremely fast compressor with lookahead, so while others would chop or ‘shape’ the wave below the threshold, it simply pushes the volume down without any perceivable pumping or other undesired side-effects. It is even good for ‘brickwalling’, if you first apply a conventional compressor to even the track’s loudness.
I can honestly say that I’ve found THE limiter that I was starving for. So thanks a lot Steve for your time and efforts! I salute you! :slight_smile:

Brilliant!

This must be included in the 2.0 release. If there is concern about “effect creep”, remove the “Hard Limiter” effect.

– Bill

This must be included in the 2.0 release. If there is concern about “effect creep”, remove the “Hard Limiter” effect.

Agree! They have different purposes though. This could be called “Soft Limiter” by contrast. :wink:
The Hard Limiter is only included in the Windows package AFAIK.

I’d prefer to just call it “Limiter…” (as now).

Although it is “soft” in the sense that it uses a soft knee (progressive compression ratio), it is “hard” in the sense that it provides infinite compression (brick wall) at the “Limit” level. In contrast the “Hard Limiter” produces hard clipping (a type of “wave shaping”), which can then be moderated by adding some of the original peaks back on (increasing the “Residual level” above zero) and / or mixing in part of the unprocessed audio (reducing the “Wet level”). Unfortunately, when softening the clipping with either of these methods, the output will probably exceed the “dB limit”.

The thing that would be lost by replacing the Hard Limiter is the ability to “hard clip”, though due to the distortion that hard clipping produces I’d say that this is not usually very useful. Where wave shaping is required, the sclimiter and Broadcast limiters will generally produce superior results than the Hard Limiter.

Where “conventional” limiting (fast compression) is required, this Limiter is the right tool and I’m not aware of any equivalent alternatives in Audacity.

Personally I don’t think the names are really important as long as everything is properly documented. That’s what’s wrong, IMHO, with Audacity’s “Leveler”, a strange compressor that adds noticeable distortion without anyone knowing exactly what it does or how it works.

The thing that would be lost by replacing the Hard Limiter is the ability to “hard clip”, though due to the distortion that hard clipping produces I’d say that this is not usually very useful. Where wave shaping is required, the sclimiter and Broadcast limiters will generally produce superior results than the Hard Limiter.

Agree. I don’t think the Hard Limiter is much better than amplifying beyond the 0dB limit. The distortion is there anyway, so maybe it could be replaced by the sclimiter. You can get proper brickwalling with it without the ugly distortion of hard clipping. This Limiter should also be added, and I don’t think including 2 different limiters is a problem, since the developers find it OK to bundle 2 compressors (I’m counting the Leveler here).

This limiter is a very handy downward-compression effect.
I was using SC4 on a reversed version of tracks to achieve a similar effect, using this plugin this is easier.

Possible bug : applied to tracks with sample rates below 44100 it changes the speed, [when used in Audacity 1.3.12 on Linux]. A workaround is to resample the track at 44100 or higher.

Thanks Trebor - good catch.
It was a bug - now fixed.
limiter.ny (1.3 KB)

The limiter adds bite …

I just installed this to attempt to tame some loud applause in a relatively quiet acoustic recording, but either I’m doing something wrong or something isn’t working right.

The level on the track is pretty low - music peaks around -17 dB and applause peaks around -11 dB. What I’d like to do is just barely touch the music peaks while bringing the applause down to that level, so I can then bring everything up without the applause causing clipping. However, stuff well below the peak is getting pushed way down. Ex:

I’m looking at 5 seconds of audio now, peak is -19.6 dB. In theory, running the limiter with “Limit to” set to -15 dB should cause the audio to barely be touched, if at all. However, if I run it (with Make-up Gain turned off), everything gets pushed down to -29.1 dB. Am I not using it as intended? “As peaks in the original audio approach this level” - how closely is “approach”?

You’re using it beyond its design limits.

The “Limit to (dB)” default setting is -3.0, and the range of the slider is -10 to 0 dB. The ends of the scale are “extreme” settings and you are going beyond the extreme low setting of -10 dB.

It’s (automatically) variable and depends on the threshold (“Limit to (dB)”) level.
When the threshold is close to 0 dB, the “knee” is quite “hard” - that is, there is no effect on audio until it gets really close to the threshold (fractions of 1 dB). With lower threshold settings the knee gets “softer” - that is, the effect is more progressive so the plug-in will start to compress at a lower level (relative to the threshold level).

A picture from Wilipedia illustrating “hard knee” and “soft knee” (Dynamic range compression - Wikipedia )

This is the intended behaviour and produces much less noticeable distortion than using a fixed knee.

Let’s say that you have the threshold set to -15 dB ("Limit to -15 dB):
If there is a peak in the audio that has a level of 0 dB, the plug-in has to reduce that peak by 15 dB so that its new level is -15 dB. With a hard knee, there would be no effect at all on audio below -15 dB, then for all sounds above this level, the level would be pulled down to -15 dB. Suddenly you are going from normal dynamics to no dynamics - whether the input sound has a peak level of -14 dB, or 0 dB (which is a BIG range), the output level would be -15 dB. The effect would be very noticeable and not very pleasant.

In this illustration I have applied the Limiter to a “ramp” (fade-in from silence to 0 dB). The first track is limited to -1 dB, the second to -3 dB and the third to -6 dB.
trackpanel000.png
You can see that with a threshold of -1 dB, the effect is almost “brick wall” - no effect until almost at the limit, then a high compression ratio to bring the maximum peak down to -1 dB.

When the limit level is set at -3 dB, the effect starts to kick in at about -4.5 dB and the compression ratio gradually increases the higher the input goes above this level. There is still some dynamic difference for high level peaks, but above -4.5 dB the dynamic range is increasingly reduced such that peaks will never exceed -3 dB.

When the limit level is set at -6 dB, you will see that the knee is soft and compression begins very gradually from a low level, gradually increasing to a high compression ratio as the level approaches the -6 dB limit.

As it says in this article: SOS Past Articles now online (back to January 1994)

A basic compressor does nothing to the input signal until it reaches the threshold, then the full amount of gain reduction is applied as fast as the attack time will let it. This is good for assertive level control, but can be a little too obvious when a lot of compression is being applied to critical sounds within a mix — or to complete mixes for that matter. A gentler-sounding compression can be achieved by using a so-called soft-knee compressor, where the compression ratio increases gradually as the signal approaches the threshold. Once the signal passes the threshold, the full ratio as set by the user is applied, but, because some compression is applied to signals approaching the threshold, the transition from no gain reduction to full gain reduction is far smoother.

In the case of this limiter, the attack and release are fixed and very fast, so if a hard knee is used with low threshold, then the compression ratio will need to be extremely high and it will be very noticeable as the limiter kicks in and out. When the threshold is not so low it should be only exceptional peaks that are affected, so we can get away with a high compression ratio and a hard knee.

In normal operation you would apply this affect after normalizing the audio to 0 dB, then set the limit level to cut down peaks that are higher than the rest of the audio.

Hopefully that explains what is happening with the Limiter, so now onto your job in hand:

The “Pop Mute” effect may be better suited to this job.
Here’s the full forum topic about Pop Mute: New plug-in effect - PopMute
Here is a direct link to the current version of the plug-in: http://forum.audacityteam.org/download/file.php?id=4086
There is a “Help Screen” included in the plug-in that explains the basic operation, but there is a lot of additional information in the forum topic.
This effect would normally be applied only to the region that you want to affect and not the entire track.
You will probably want the “Mute Level” set quite high so that the clapping does not disappear altogether.

Thanks. I’ll take a look later today.

In my case, since the claps are very transient (rather than an overall high level “roar”), I’m not too concerned about a hard knee. I’ve played around with a regular compressor, but since there’s no look ahead, the claps are still an issue even with a very fast attack time.

The same three LADSPA-ported effects that are included in Audacity for Windows (Hard Limiter, GVerb and SC4) are also in Audacity for Mac.


Gale

Yo hope I’m not disrupting anything bringing this topic out of the grave, but I was wondering how you actually apply the ‘limit’ on tracks? I’ve downloaded and properly installed the Limiter, I can use it afterwards as an effect, but how do you get the tracks to have 'db limit’s like that?

I would not recommend the LADSPA “Hard Limiter” unless you intend to create distortion.
As a (non-distorting) peal limiter effect the “Limiter (2)” plugin is much better. It may be downloaded here: Missing features - Audacity Support
Instructions to install it: Missing features - Audacity Support

As with other effects in Audacity, to apply the effect, select the audio that you wish to apply it to, then go the the “Effect” menu and open the required effect.
Audacity cannot apply effects during recording. Effects are always applied to selected audio in existing audio tracks.