EQ: Linear Phase vs Minimum Phase

[Music] hi and welcome to this tutorial about linear phase in EQ fabfilter pro-q plug-in offers a selection of linear phase modes with progressively higher latencies it would be reasonable to assume that the default minimum phase mode is the lowest quality setting and the setting with the highest latency provides the highest quality it's not as simple as that however while the hi and Max latency linear phase modes do indeed provide better quality than the lower latency linear phase modes in most situations the default zero latency mode will actually perform better than any other let's start with

a kick drum which have created using twin 2 running through Saturn and which have sculpted heavily with EQ working from the left of a high-pass filter to remove the very lowest frequencies but also with some resonance to create a boost at about 50 Hertz then a steep sharp notch filter just below 90 Hertz to make room for some important frequencies in the bass guitar part plus a wide deep cut at 500 Hertz to remove some stuff I just don't like let's solo the kick channel and switch to the maximum latency linear phase mode and notice

the difference this is made to the initial transient instead of starting with a solid clicky attack each hit is now preceded by a short crescendo of the low boomy part of the sound almost like a reverse reverb effect if I switch back to minimum phase zero latency mode each kick drum hit now starts with a clicky attack again and the low boom follows afterwards as it should so what's actually going on here I can demonstrate it visually using a special impulse signal known as a Dirac spike which is simply one single full-scale sample in a

file full of zeros if I process this with a low-pass filter in zero latency mode and render the result we can see that the filter rings for a short period after the original spike processing again in linear phase mode produces a similar ringing effect but this time it starts before the original transient and continues afterwards this might seem to be impossible until you remember that the linear phase modes all introduce an overall delay to the signal which has been compensated for automatically by the da W it is this pre ringing that accounts for the transients

smearing effect in my kick drum example pre ringing tends to be much more noticeable than the post ringing effects caused by minimum phase filtering and is a general problem with all linear phase algorithms not just those in pro-q I confess I've manufactured this example deliberately to make the pre ringing as obvious as possible by applying fairly drastic low-frequency eq shaping and by choosing a source that combines lots of low frequency energy with a hard transient attack in most cases pre ringing will be much more subtle in this so we have an example of linear phase

mode definitely not being an improvement over zero latency mode but in what situation would linear phase be useful well perhaps I should first of all clarify what we mean by the term phase in the first place let's take a simple sine wave the basic building block of all sounds just one pure frequency in this case at 200 Hertz and I'll duplicate the track to create an exact copy adding these together results much as you would expect in an identical sine wave with twice the amplitude equivalent to a gain boost of 6 dB now notice what

happens if I slide the copy later the resulting sine wave gets quieter and quieter until we reach the point where the peaks in the copied waveform line up with the troughs in the original and the two cancel out completely by changing the timing of the second waveform I've changed its phase relative to the first waveform phase being the term we use to describe how far away furs traveled through its cycle if I continue to slide the copy back in time the two sine waves will stop cancelling out and will eventually come back full circle with

both waves in phase with each other again like a circle we measure phase in degrees so when two waves are equal and opposite to each other we say that they are 180 degrees out of phase while 360 degrees of phase shift puts us back where we started with both waves perfectly in phase again of course another way to create a phase shift of 180 degrees would be to invert the polarity of the wave so that positive cycles become negative and vice versa and our two sine waves now cancel out again so we can see that

when mixing two identical or very similar waves together the phase relationship between the two waves will have a profound effect on the result a commonly cited example of this would be a drumkit recording using a multiple microphones the example you see here is a short multitrack drum loop played by Stefan o Esposito and recorded with a total of eight mics when considering the sound of the snare drum we therefore have to take into account the contributions from both top and bottom snare mics the overheads and the single room mic and also the phase relationship between

them if I solo the snare close mics individually we hear that they sound quite different but being two mics pointing at the same snare drum there are also many similarities between the two or in other words the two signals are correlated if I solo both mics together and invert the polarity of just the top snare using the phase button on the mixer channel the combined sound changes noticeably and the low fundamental of the drum is much reduced in level so let's put the mics back in phase with each other add a pro cue to the

top snare channel and try applying some typical EQ shaping perhaps to address the ringing at around 400 Hertz and actually it works so much as you'd expect boosting that frequency makes the ringing louder and more prominent while cutting chains the ringing and makes the snare sound tighter and more damped if I switch to linear phase the difference is very subtle let's load Pro Q into an analyzer to try to see what's going on this is Christian buds free VST plug-in analyzer software and I'm going to choose to measure the plugins frequency response this produces a

graph much like the one on Pro Qzone interface and any changes I make to the EQ settings are mirrored by the analyzer after a short delay the analyzer can also be set to measure phase response however via the domain menu at the top and the graph now displays the phase change is caused by the EQ cut I dialed in as expected switching to a linear phase mode results in a perfectly flat line indicating no phase shifts at any frequency but there are some interesting points to note about the phase shifts caused back in minimum phase

mode these shifts are both positive and negative and they vary in proportion to the amount of gain I dial in for the band if I boost instead they swap over but notice that the center frequency I'm targeting with the EQ remains at zero phase all the time in fact the biggest phase shifts happen when the gradient is steepest on the EQ curve not at the frequency with the greatest cut or boost if I switch to a low shelving filter instead this time we have a single phase shift again centered on the region with the steepest

gradient and the phase trends back towards zero in the flat regions above and below the shelf so I'm now going to analyze the frequency response again but this time for a parallel signal chain in which the output from the EQ is added back together with the original signal with no EQ applied we simply have a doubling of amplitude so an overall volume boost of about 6 dB and applying an EQ cut results in a slight dip at that frequency as I cut it out of just one half of the signal path let's switch to linear

phase and notice that the overall amount of cut applied doesn't change as our target frequency is at zero phase in either case instead the differences occur where the phase changes are greatest above and below the target frequency so the overall result is just a narrowing of the bell shape quite a subtle difference which can be almost completely dialed out by adjusting the cue to compensate likewise if I switch to EQ shape to a low shelf again the overall effect as much as you'd expect and the difference between linear phase and minimum phase amounts to just

a slight difference in the slope and the frequency of the shelf so what's going to happen if I choose a high-pass filter instead well in linear phase mode the results are again much as you'd expect the signal starts at unity coming only from the dry signal path then the level rises smoothly as the high pass signal starts to contribute with an overall shape like a high shelving booster if I switch to minimum phase however we now see quite a dramatic difference with the steep knotch suddenly appearing near the cutoff frequency if we look at the

phase response of the high-pass filter we can see why unlike the subtle gentle phase shifts from the bell and shelving filters the high-pass filter has a dramatic effect on the phase with a four hundred and eighty degree phase reversal at the cutoff fruits so let's go back to our snare drum example still listening to a mix of both top and bottom snare mics and dial in a high-pass filter for the top mic instead of a bell sweeping the filter up from the low frequencies we start to hear the low fundamental of the drum drop in

level even though we've not reached high enough to be reducing the frequency with the filter at all if I solo the top snare on its own and toggle the filter in and out we don't really hear any difference but with both mics on there is definitely a difference this is of course because the high-pass filter has applied almost exactly a hundred and eighty degrees of phase shift to those frequencies and therefore changed the way the two signals mixed together I can prove this by reversing the polarity of the top snare using the channel phase button

again and unlike before the low fundamental of the drum is now much stronger with the top mic flip switching to a steeper filter will result in even more unpredictable side effects with a 48 DB per octave filter set a little below 180 her's we find the ringing frequency we identified earlier has suddenly disappeared completely if i bypass the filter it comes back filter bucking and it's gone again if I switch to linear phase mode however this ringing frequency is not affected the filter cuts out the low base without affecting the way the to snare mics

mixed together up to the analyzer again and we can clearly see the massive notch in the frequency response that's killing the snare drum ring but in linear phase mode the dry and wet signals combined together with no unintended side effects it's still not entirely clear which mode is better however even in this example linear phase mode produces more predictable results by avoiding changing the phase relationship between correlated signals but changing the phase relationship between correlated signals is not necessarily a bad thing and in some cases might help them fit together in subjectively more pleasing ways

i'm going to manufacture a more extreme example for you let's root to kick and the bass parts from my earlier example to another channel then low-pass filter this channel down to about 100 Hertz to isolate the low bass and add a pro C set to compress the results quite hard mixing this signal in under the rest of the mix should create a frequency specific upward compression effect like the one I used for the high frequencies in the mastering tutorial series to glue the kick and bass parts together and to add weight to the whole mix

however while adding in this channel seems to raise the level of the kick in the mix it's not really helping the bass tool [Music] until I switched a linear phase mode [Music] this time the difference is pretty clear linear phase mode sounds much better than minimum phase in this situation again the analyzer shows us why in a parallel configuration the phase shift from the minimum phase filter causes a steep notch to appear near the cutoff frequency and this was a critical region for the bass guitar sound no such a notch appears in linear phase mode

however minimum phase filters still have some uses even in this setup I'll turn down the cue for the filter to produce a more gentle slope and notice that the dip in the overall frequency response is now much reduced in fact it begins to resemble the Gers on shape which combines a shelving boost with a cut just above it and which can sometimes sound better than just a simple shelf EQ [Music] so let's June the cutoff frequency higher to about 500 Hertz the frequency response now dips around 500 Hertz a region that's not that important to

either the kick or the bass sounds in this case and then Rises smoothly in the low end switching back to our kick and bass example it's no longer obvious that linear phase is better [Music] in fact I might now prefer the results in minimum phase mode of course it's also worth considering the fact that I'm running a compressor on the signal so the level of the filtered signal will also be changing relative to the dry signal the easiest way to see this in the analyzer is to adjust the level of the wet signal using pro

Qzone output level knob when the kick and bass levels are high the compressor will be reducing the gain of the filtered signal so the overall frequency response will flatten out then when the compressor releases the levels will increase again so boosting the low frequencies and cutting the mids in other words we've created an upward compression effect for the low frequencies and also a downward expansion effect for the mid frequencies above it in linear phase mode however there is no downward expansion of the mids we are now just purely compressing the low-end so in summary most

of the time for conventional EQ in Duty's the minimum phase mode will be the optimum choice as well as being conveniently free of latency the linear phase modes are more specialized options that will behave more predictably and intuitively when mixing correlated signals but might not necessarily sound better regarding the different latency options most da WS will compensate for latency when mixing so you can safely choose the better quality high and Max settings when you need a linear phase option if latency is an issue try progressively lower settings and listen out for any unwanted artifacts especially

when dealing with low frequency content that's all I have time for in this video thanks for watching [Music] you [Music]