Eriksimon wrote:
That has to do with the Q value and how Low cut and/or High Pass filters behave: if you have a very steep Q setting, there will be boosting "around the edges".
You are right that this can be an issue, but it doesn't have to be the cause here. There's something more subtle and counter intuitive that can change the signal level even for an amplitude preserving allpass filter.
The problem is how the level is measured. In a DAW the most important kind of level measure is simply peak amplitude, making it easy to avoid clipping. Unfortunately peak amplitude is not a well behaved or even conserved measure when using linear filters like an allpass or the low cut used by the OP.
If you had a single sinusoid with constant frequency, then the frequency response of the filter would tell you exactly how its amplitude changes upon filtering. However, if you have a mixture of sinusoids like in any real world signal, the peak amplitude of those does not follow directly from the magnitude response graph, because their phase relationship is also altered by the filter.
We can even build a filter that has a unit magnitude response, a so called allpass filter, and it is capable of changing the phase relationship of a signal in a way that influences the peak level reading in either direction. You can avoid this effect by restricting yourself to the use of linear-phase filters, which don't introduce relative phase changes. This restriction may be undesirable however, because those filters have quite a few disadvantages too.
The other way to avoid such confusing effects is to simply rely on the RMS measure instead of the peak level meter. The RMS measure is strictly preserved by allpass filtering and in general the RMS measure of the filtered signal is bounded by the RMS measure of the original signal multiplied by the RMS measure of the filter response.
So as you understand now, there's nothing broken and you can just turn the fader down a bit and everything will be fine.
Cheers,
Jazz
