I was wondering - do they shift the phase? Or not?
In EQs, high pass filters do shift the phase, which usually results in the signal getting louder, even though you are cutting away from it. (Except for linear phase EQs, of course.) That would be a problem with compressors, as they are supposed to accurately detect loudness and then reduce it.
Has this (potential? IDK!) issue been solved in compressor-land, or is it just being glossed over/ignored, hoping that noone would notice?
HPFs in compressor detector circuits
-
- KVRAF
- 2476 posts since 15 Apr, 2004 from Capital City, UK
I'm pretty sure the energy of a signal isn't affected that much buy phase shift, on it's own.
Since a detector circuit is only listening, and isn't actually mixed into the output at all, and the detector only cares about the volume of the signal (however that might be measured.. RMS, Peak..) any phase shifting of that signal really won't affect the way the gain circuit works.
Maybe if you're using steep linear phase filters, which would, as you know, add pre-ringing (and latency) that additional 'added' signal could well affect the detector, but not in any significant ways that would be detrimental to the function of the compressor.
Ultimately, and I will be corrected by this community if I'm wrong, it's a non-issue as far as most compressor consumers and most devs are concerned, I would have thought.
Since a detector circuit is only listening, and isn't actually mixed into the output at all, and the detector only cares about the volume of the signal (however that might be measured.. RMS, Peak..) any phase shifting of that signal really won't affect the way the gain circuit works.
Maybe if you're using steep linear phase filters, which would, as you know, add pre-ringing (and latency) that additional 'added' signal could well affect the detector, but not in any significant ways that would be detrimental to the function of the compressor.
Ultimately, and I will be corrected by this community if I'm wrong, it's a non-issue as far as most compressor consumers and most devs are concerned, I would have thought.
-
- Beware the Quoth
- 33210 posts since 4 Sep, 2001 from R'lyeh Oceanic Amusement Park and Funfair
-
- KVRAF
- 4434 posts since 15 Nov, 2006 from Hell
I imagine it would be trivial to test by taking any compressor with full band side chain, feeding minimum/linear phase high passed sounds into it, and comparing results for both. I imagine the differences will be negligible, if at all measurable.
I don't know what to write here that won't be censored, as I can only speak in profanity.
-
- KVRian
- 779 posts since 21 Sep, 2008
Yes, most HPFs in compressors shift phase. And indeed, if the incoming signal is already squashed this would typically increase peak level. However, it is not that linear phase EQs are totally free of the same phenomemon (look for Gibbs effect).
Using linear phase filter in the side chain may also result in the unwanted behavior due to pre-ringing which is produced by the linear phase filters. If the threshold of the dynamic processor is suffiently low, pre-ringing will cause it to react before the actual peak comes. Although, this effect should be minor.
-
- KVRist
- 162 posts since 1 Jul, 2020
I would say yes they do shift phase of the detector signal and likely cause an increase in gain...
This would mean your threshold setting would need adjusting to compensate.
BUT if you think about what happens if you're detecting a high passed version, chances are signal is now lower due to loss of low frequency. So it likely balances itself out.
Overall nothing to worry about. We all adjust settings until it sounds right. No need to over think
This would mean your threshold setting would need adjusting to compensate.
BUT if you think about what happens if you're detecting a high passed version, chances are signal is now lower due to loss of low frequency. So it likely balances itself out.
Overall nothing to worry about. We all adjust settings until it sounds right. No need to over think
-
Andreya_Autumn Andreya_Autumn https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=553235
- KVRist
- 237 posts since 21 Feb, 2022
Phase shift yes, certainly If that were a linear phase high pass in there, it would add latency and pre-ringing. Both of which would usually be more problematic than the phase shift.
And the increase in gain depends what you mean by gain. If you feed in a low freq square wave, you could indeed see higher peak-to-peak levels from phase shifting the partials. If the compressor is RMS-sensing, that won't make any difference, as the average level is still the same, though if it's peak-sensing the phase shift could maybe trigger more compression.
But for almost any other situation (say a full mix), it's about equally likely you *decrease* the peak level as increase it.
And indeed the reason those high pass filters are there in the first place is to prevent all that bass energy from taking up all of the compressors attention. In pretty much every case the overall level of the sidechain signal lessens when these are used.
The phase shift can interact with the attack/knee settings to shape the groove though! Molot GE from TDR has a switch to throw some all pass filters in the sidchain path, which is admittedly a pretty subtle effect, but pretty cool sometimes on bass-heavy parts. You can use it with the high pass off too.
And the increase in gain depends what you mean by gain. If you feed in a low freq square wave, you could indeed see higher peak-to-peak levels from phase shifting the partials. If the compressor is RMS-sensing, that won't make any difference, as the average level is still the same, though if it's peak-sensing the phase shift could maybe trigger more compression.
But for almost any other situation (say a full mix), it's about equally likely you *decrease* the peak level as increase it.
And indeed the reason those high pass filters are there in the first place is to prevent all that bass energy from taking up all of the compressors attention. In pretty much every case the overall level of the sidechain signal lessens when these are used.
The phase shift can interact with the attack/knee settings to shape the groove though! Molot GE from TDR has a switch to throw some all pass filters in the sidchain path, which is admittedly a pretty subtle effect, but pretty cool sometimes on bass-heavy parts. You can use it with the high pass off too.
-
- KVRian
- 919 posts since 4 Jan, 2007
Linear phase might not be needed, the dry signal might also be passed through an allpass filter with a matching phase shift.
In e.g. Presswerk there is a phase rotator for the key signal, but that is more heavy than a 1-2 pole high pass.
Normally one doesn't have new ideas or notice new problems in established fields that one is not expert in. Lots of clever people have spent lots of man years in every established field already.
In e.g. Presswerk there is a phase rotator for the key signal, but that is more heavy than a 1-2 pole high pass.
Normally one doesn't have new ideas or notice new problems in established fields that one is not expert in. Lots of clever people have spent lots of man years in every established field already.
Last edited by rafa1981 on Sat Apr 20, 2024 12:49 pm, edited 1 time in total.
