Sharing Mastering Knowledge

In the coming weeks, the usual processors involved in the mastering chain will be reviewed. Since it's last in this chain, the limiter will be the first to be considered.



The limiter is supposed to be the very last treatment in the mastering process with an analog counterpart. The sample rate conversion, the bit depth conversion and associated dithering and noise shaping will be examined in future topics.
This short video demonstrates that a limiter prevents audio from clipping. The process is a speciatized compressor featuring both zero attack time and infinite ratio. It's a key processor when dealing with high SPL in the digital domain.

The limiter main controls are threshold and release time. Unlike a regular compressor which refers to the signal average value (RMS), the limiter threshold value refers to the instant peak value. The limiter action is driven by signal peaks since it has to prohibit any violation of the threshold level. To prevent audible distorsion, the gain variation has to be released. Short release values increase the distortion, and long ones create an unpleasant pumping effect. That's why most limiters feature an auto-release mode keeping distortion low while preserving from pumping artefacts.

The video shows the action of a soft clipper. This particular limiting has no release time. With a properly set threshold, it only affects fast transients and keeps most of the signal untouched. There is no gain envelope here. This process is based on a transfer curve (in/out) shaping. The top of this curve is bended in such a manner that the input signal above the threshold never reaches the le maximum output value (0dB fs). This method produces a highly distorted signal when acting. But it can be pleasant for managing rare ~1dB peaks on percussive signal. In all cases it is preferable to hard clipping because it produces less random harmonics.

The release time management is the key point to prevent distorsion and pumping artifacts when limiting. The auto-release mode must be used when available. It's aways better than a constant release value. It produces a less noticable limiting. The auto-release quality must determine your limiter choice. By controlling the gain envelope, this function also determines the quantity and the quality of the distorsion introduced by the gain variations of the processing. If all limiters are supposed to have a linear frequency response, the fast gain variations affect this linearity by modulating the audio signal. That's why every auto-release design has its own colour.