FabFilter User Forum

L2 limiting below 0dB in Oversampling mode

Hi,

while analyzing and comparing some different EDM tracks, i found something that I do not really understand:

One track (which I chose for this because it feels very loud and well mixed compared to almost everything else I know) triggers L2's limiting, even with 0.0dB input gain. The Track is peaking at -0.1dB. True peak is over +1dB, however the true peak function of L2 was not engaged. (When engaged the Limiter is engaged even more)
This happens only when Oversampling is engaged, and leads to L2 taking off over 1,4dB of volume.
This is not the case for other tracks of the same genre (also with positive true peak levels), or only to a much smaller degree.

So I guess my question is: what could be the difference in this track that makes it peak so much higher with oversampling engaged, while peaking at the same level as others without it?
And also: Is this something to worry about during live Playback, or is it rather a "hack" someone found, that lets him get a track that feels much louder, without clipping when played back in the original sample rate?

Thank you very much in advance!

Robin

Hi Robin,

Oversampling increases the precision of the peak detection circuit. In fact, one of the main processing of calculating true peak levels is actually oversampling the signal. There's a few more steps involved in calculating true peaks, but matching the oversampling will already get you very close to a true peak measurement.

When comparing mastered material there are lots of variables, like whether they used true peak limiting when mastering, whether they used oversampling and how much, etc.

It's possible the track that you measured did not use true peak limiting or oversampling during mastering to keep its loud character. Some mastering engineers like to keep true peak limiting disabled to keep an extra dB or two of dynamic range, which as a compromise will be above true peak. It's a choice between loudness or distortion.

The other tracks may have used oversampling or true peak limiting, which would cause the difference.

Ralph (FabFilter)

Thanks for the detailed answer Ralph!

Robin

Reply to this topic: