Does limiting make your audio clip and distort?

Push your levels too high and you will break the bounds of your system's capability. But will a limiter make things any better, or simply act as another source of distortion?

Here's an experiment... Take a section of cleanly recorded audio that is well below the level at which the red lights come on.

Now raise the level, raise it some more, then raise it some more. Your system almost certainly does have the ability to raise the level by as much as you want, although you may have to rummage among your plug-ins to find it.

But by whatever means, raise the level until the red light starts to flicker. Listen carefully for changes in the sound. Continue raising the level until the red light is solidly on and there is absolutely no doubt that your sound is distorted.

What you are listening to is the sound of clipping. Any digital audio system has an upper level beyond which the signal just cannot go. We call this 0 dBFS - zero decibels with respect to full scale.

If you try to raise the signal beyond this level, the tips will be clipped, causing distortion.

By the way, this test should be done at a low monitoring level as it will produce a lot of high frequency energy that could potentially damage your tweeters.

To avoid clipping, you might consider using a limiter. Set correctly, this will prevent the signal level reaching 0 dBFS, so no clipping can occur.

Well, that is the theory. In practice it depends on making the correct settings.

Think of it like this... imagine that you could control the level by hand very quickly so that the signal never went above 0 dBFS. Every time a loud section comes along, you lower the level so that the signal stays within bounds. When it goes quiet again, you can raise the level back up again.

That's exactly what a limiter does. But you can make a limiter respond very much more quickly.

You might therefore be tempted to set a really short 'attack' time - that's the time it takes for the limiter to respond; it has nothing to do with making your signal sound more attacking subjectively.

But when the speed of response approaches 50 milliseconds or less, something interesting happens. The limiter is now fast enough to respond to single cycles of the waveform, at low frequency.

At this point, the limiter has the capability of changing the shape of single cycles of the waveform.

There is a word for that - distortion.

So you have a dilemma. Set the attack time too long and fast peaks may get through and clip. But set it too short and the limiter itself causes distortion. It isn't clipping, but it can sound as bad.

The moral is to be very careful when setting an attack time of less than around 100 milliseconds, to be on the safe side. Listen very carefully, watch for red lights, and perhaps limit a little lower than 0 dBFS, then normalize afterwards to bring the level up to peak.