Login / Register
 
Cutting frequencies actually INCREASES overall volume of pre-master??
New Reply
Subscribe
Rdodz
Thread Starter
#1
18th April 2012
Old 18th April 2012
  #1
Gear Head
 
Joined: Jun 2010
Posts: 70

Thread Starter
Rdodz is offline
Cutting frequencies actually INCREASES overall volume of pre-master??

I'm mastering a track, and rolling off at 25hz

EQ (Fab Filter Pro-Q)
Compression (Waves SSL)
Limiter (Voxengo Elephant)


When I don't cut under 25hz: the limiter is only taking off -3dB gain reduction

When I cut under 25hz the limiter has to work harder (?!!!!???!): it's taking off -5.6dB of gain reduction

WHY?

I can't explain this. I'm getting rid of unwanted frequencies but yet it seems that this increases volume of my track?
#2
18th April 2012
Old 18th April 2012
  #2
Lives for gear
 
yllet's Avatar
 
Joined: Jun 2008
Location: Malmö. Sweden

yllet is online now
Quote:
Originally Posted by Rdodz View Post
I'm mastering a track, and rolling off at 25hz

EQ (Fab Filter Pro-Q)
Compression (Waves SSL)
Limiter (Voxengo Elephant)


When I don't cut under 25hz: the limiter is only taking off -3dB gain reduction

When I cut under 25hz the limiter has to work harder (?!!!!???!): it's taking off -5.6dB of gain reduction

WHY?

I can't explain this. I'm getting rid of unwanted frequencies but yet it seems that this increases volume of my track?
Not that strange really... When you cut some frequencies others become stronger, that's the way it works. In this case, more "hi-end peaks" gets pronounced and therefore triggers the limiter.

BTW are you using a really steep filter when u cut? Like 36-48db, thats a no no in mastering for me, at least 6-12db should do the trick if needed.

All the best from Sweden

/Jon
__________________
It's not about the gear, but what you hear...
http://jon.rinneby.se/
http://turnoffyourtelevision.se
Trakworx
Verified Member
#3
18th April 2012
Old 18th April 2012
  #3
Lives for gear
 
Trakworx's Avatar
 
Joined: Apr 2008
Location: San Francisco, CA
Posts: 2,595
My Recordings/Credits

Verified Member
Trakworx is offline
Quote:
Originally Posted by Rdodz View Post
I'm mastering a track, and rolling off at 25hz

EQ (Fab Filter Pro-Q)
Compression (Waves SSL)
Limiter (Voxengo Elephant)


When I don't cut under 25hz: the limiter is only taking off -3dB gain reduction

When I cut under 25hz the limiter has to work harder (?!!!!???!): it's taking off -5.6dB of gain reduction

WHY?

I can't explain this. I'm getting rid of unwanted frequencies but yet it seems that this increases volume of my track?
Try bypassing the Waves SSL compressor and see what happens.
#4
18th April 2012
Old 18th April 2012
  #4
mymixisbetterthanyours!
 
Joined: Oct 2006
Location: Salzburg / Berlin
Posts: 2,435

kosmokrator is offline
Quote:
Originally Posted by Rdodz View Post
I'm mastering a track, and rolling off at 25hz

EQ (Fab Filter Pro-Q)
Compression (Waves SSL)
Limiter (Voxengo Elephant)


When I don't cut under 25hz: the limiter is only taking off -3dB gain reduction

When I cut under 25hz the limiter has to work harder (?!!!!???!): it's taking off -5.6dB of gain reduction

WHY?

I can't explain this. I'm getting rid of unwanted frequencies but yet it seems that this increases volume of my track?
Waveform reshaping.
Lagerfeldt
Verified Member
#5
18th April 2012
Old 18th April 2012
  #5
Audio Alchemist
 
Lagerfeldt's Avatar
 
Joined: Sep 2004
Location: Copenhagen, Denmark
Posts: 5,157

Verified Member
Lagerfeldt is offline
It's the phase relations that change.

The original waveform (your mix) consists of a variety of frequencies that interact and constitute a whole.

When you use an equalizer you're not only changing the amount of gain in a particular frequency area, but you're also changing the phase and therefore the timing between the various frequencies.

This causes the various frequencies to interact in a different way that may or may not lead to a positive change (peak) as the frequencies shift - even though you're cutting. Remove enough though and you'll see the peak level go down again, but that's not the solution of course.

Most equalizers are so-called minimum phase and have this natural and normally desirable behaviour. Quite often when you're equalizing you're not just changing the amount of gain in a particular area, but you're trying to obtain the right phase response.

However, you can use a phase linear equalizer to offset the relative phase changes. In your case you'll observe almost no peak change, but still get the low cut you're looking for. Linear phase is no magic bullet though, and you'll probably hear some transient blurring caused by pre-filter ringing, which is an unwanted side-effect of linear phase technology. This is one of the reasons why linear phase is not the first choice for many engineers.

Logic Pro has a bundled linear phase equalizer, but you can also get the DMG Audio EQuality and switch between modes on the fly.
__________________


Online Mastering.dk

I have come here to chew bubblegum and kick ass... and I'm all out of bubblegum
Rdodz
Thread Starter
#6
18th April 2012
Old 18th April 2012
  #6
Gear Head
 
Joined: Jun 2010
Posts: 70

Thread Starter
Rdodz is offline
so how do i get rid of unwanted ultra low frequencies without making the limiter work harder: therefore having to turn the limiter down?

All i wana do is get rid of the low nonsense, and keep the track nice and loud. But cutting the subs here is making my limiter work too hard as the track increases in volume, so i have to ease up on the gain reduction which is undesireable.
Rdodz
Thread Starter
#7
18th April 2012
Old 18th April 2012
  #7
Gear Head
 
Joined: Jun 2010
Posts: 70

Thread Starter
Rdodz is offline
Quote:
Originally Posted by Lagerfeldt View Post
It's the phase relations that change.

The original waveform (your mix) consists of a variety of frequencies that interact and constitute a whole.

When you use an equalizer you're not only changing the amount of gain in a particular frequency area, but you're also changing the phase and therefore the timing between the various frequencies.

This causes the various frequencies to interact in a different way that may or may not lead to a positive change (peak) as the frequencies shift - even though you're cutting. Remove enough though and you'll see the peak level go down again, but that's not the solution of course.

Most equalizers are so-called minimum phase and have this natural and normally desirable behaviour. Quite often when you're equalizing you're not just changing the amount of gain in a particular area, but you're trying to obtain the right phase response.

However, you can use a phase linear equalizer to offset the relative phase changes. In your case you'll observe almost no peak change. Linear phase is no magic bullet though, you'll probably hear some transient blurring caused by pre-filter ringing, which is an unwanted side-effect of linear phase technology.

Logic Pro has a bundled linear phase equalizer, but you can also get the DMG Audio EQuality and switch between modes on the fly.


Brilliant, thank you for this wonderful answer!
Lagerfeldt
Verified Member
#8
18th April 2012
Old 18th April 2012
  #8
Audio Alchemist
 
Lagerfeldt's Avatar
 
Joined: Sep 2004
Location: Copenhagen, Denmark
Posts: 5,157

Verified Member
Lagerfeldt is offline
You're welcome.
#9
18th April 2012
Old 18th April 2012
  #9
Gear addict
 
philip's Avatar
 
Joined: Feb 2004
Location: Sweden
Posts: 475

philip is online now
Quote:
Originally Posted by Rdodz View Post
so how do i get rid of unwanted ultra low frequencies without making the limiter work harder: therefore having to turn the limiter down?

All i wana do is get rid of the low nonsense, and keep the track nice and loud. But cutting the subs here is making my limiter work too hard as the track increases in volume, so i have to ease up on the gain reduction which is undesireable.
25Hz is not ultra low and it is not nonsense, in most masters it is important to preserve these frequencies. If your main speakers ain't up for the job then get a pair of headphones for checking bass.

Elephant's got a built in HP filer. Did you try that?
#10
18th April 2012
Old 18th April 2012
  #10
Gear maniac
 
Joined: Dec 2009
Location: Chicago
Posts: 212

strdsk is offline
I agree with Phillip. 25Hz is something you shouldn't be cutting unless needed...not across the board...
huejahfink
Verified Member
#11
18th April 2012
Old 18th April 2012
  #11
Lives for gear
 
huejahfink's Avatar
 
Joined: Jan 2010
Location: underneath the dank, cobbled streets of Landon Taaaan'
Posts: 1,968

Verified Member
huejahfink is offline
Quote:
Originally Posted by philip View Post
25Hz is not ultra low and it is not nonsense, in most masters it is important to preserve these frequencies.
True - not by default - but hold on -
the OP is probably not talking about cutting it completely... just reducing it.
Whether or not that frequency is nonsense / too much / etc - or not in this track, we can only guess at that without hearing it and / or speaking with the client.
__________________
Discogs Technical Credits
#12
18th April 2012
Old 18th April 2012
  #12
Gear addict
 
Joined: Oct 2009
Posts: 352

JonMiller is offline
Quote:
Originally Posted by yllet View Post
Not that strange really... When you cut some frequencies others become stronger, that's the way it works. In this case, more "hi-end peaks" gets pronounced and therefore triggers the limiter.

BTW are you using a really steep filter when u cut? Like 36-48db, thats a no no in mastering for me, at least 6-12db should do the trick if needed.

All the best from Sweden

/Jon
This is physically impossible, at least with an equalizer. Matter and Energy cannot be created nor destroyed, they can just change forms.

All things consistent, the EQ is boosting the level or you somehow broke E=MC^2
#13
18th April 2012
Old 18th April 2012
  #13
Gear addict
 
Joined: Oct 2009
Posts: 352

JonMiller is offline
Quote:
Originally Posted by Rdodz View Post
so how do i get rid of unwanted ultra low frequencies without making the limiter work harder: therefore having to turn the limiter down?

All i wana do is get rid of the low nonsense, and keep the track nice and loud. But cutting the subs here is making my limiter work too hard as the track increases in volume, so i have to ease up on the gain reduction which is undesireable.
Drop the output of the EQ 2db to compensate.
#14
18th April 2012
Old 18th April 2012
  #14
Lives for gear
 
Quint's Avatar
 
Joined: Apr 2008
Location: Austin, TX
Posts: 2,003

Quint is offline
Quote:
Originally Posted by Trakworx View Post
Try bypassing the Waves SSL compressor and see what happens.
I agree with this. It's possible that the way you have your SSL setup was causing it to clamp down a lot harder when the louder bass frequencies were present. Once you lowered the bass frequency's volume, the SSL quit clamping down as hard and a higher volume level started to make it's way to the limiter, causing it to clamp down harder to make up for lighter amount of gain reduction that the SSL is now contributing.
__________________
We (AudioLot) accept VISA/MC, Paypal, Google Checkout, Bank Wire, Checks, and Cash. - Kittonian
Greg Reierson
Verified Member
#15
18th April 2012
Old 18th April 2012
  #15
Lives for gear
 
Greg Reierson's Avatar
 
Joined: Jan 2007
Location: Minneapolis
Posts: 1,731

Verified Member
Greg Reierson is offline
Quote:
Originally Posted by JonMiller View Post
This is physically impossible, at least with an equalizer.
Filters have ring and that resonance can spike above the level of the original signal.

Back to the OP: I've seen the same thing happen when removing excessive <30-40Hz from kick drums, etc.

GR
__________________
Greg Reierson
www.RareFormMastering.com
Vinyl : CD : Analog : Digital

"If the album isn't competing for singles airplay in focus groups, forget about loudness!" - Bob Olhsson
Lagerfeldt
Verified Member
#16
18th April 2012
Old 18th April 2012
  #16
Audio Alchemist
 
Lagerfeldt's Avatar
 
Joined: Sep 2004
Location: Copenhagen, Denmark
Posts: 5,157

Verified Member
Lagerfeldt is offline
It's very simple to test by performing the minimum phase versus linear phase change I suggested above in my explanation. I doubt it's related to the SSL, but that's easily tested by leaving the SSL out of the chain.

The sudden change in peak level isn't caused very much by the filter ringing (though it does contribute and especially with steep slopes) but usually by the phase changes alone in these cases.

Try switching from a 12 dB/Oct minimum phase low cut at e.g. 30 Hz on a heavily compressed and limited kick drum to a ditto linear phase. The change in peak level will be minimal with linear phase, and the remaining peak level change is caused by the filter ringing. Usually less than 0.3 dB compared to anything from 0.5 dB to 3 dB caused by the phase change itself in minimum phase mode.
Greg Reierson
Verified Member
#17
18th April 2012
Old 18th April 2012
  #17
Lives for gear
 
Greg Reierson's Avatar
 
Joined: Jan 2007
Location: Minneapolis
Posts: 1,731

Verified Member
Greg Reierson is offline
Yep, that makes sense.

GR
#18
18th April 2012
Old 18th April 2012
  #18
Gear addict
 
Joined: Oct 2009
Posts: 352

JonMiller is offline
Quote:
Originally Posted by Greg Reierson View Post
Filters have ring and that resonance can spike above the level of the original signal.

Back to the OP: I've seen the same thing happen when removing excessive <30-40Hz from kick drums, etc.

GR
How can that occur in a cut with a High Pass Filter? I could see with a steep notch, but not a true High Pass
Alexey Lukin
Verified Member
#19
18th April 2012
Old 18th April 2012
  #19
Lives for gear
 
Alexey Lukin's Avatar
 
Joined: Dec 2007
Posts: 959

Verified Member
Alexey Lukin is offline
The topic title is misleading: the EQ that only cuts cannot increase your "volume". Not even your RMS level. What you see is increase in peak levels.

It happens because* of filter ringing (oscillation of filter's impulse response), which is higher in nonlinear-phase filters, but is still present in linear-phase filters.

Solution — apply a limiter after your filter, use a linear-phase filter if possible.

_______________________________
* On a technical side, ringing (more specifically — negative overshoot in the filter's impulse response) is the only direct cause of peak levels increase. Of course, ringing, in its turn, is totally dependent of steepness of the filter's frequency response and linearity of its phase response, but those are already indirect factors.
Lagerfeldt
Verified Member
#20
19th April 2012
Old 19th April 2012
  #20
Audio Alchemist
 
Lagerfeldt's Avatar
 
Joined: Sep 2004
Location: Copenhagen, Denmark
Posts: 5,157

Verified Member
Lagerfeldt is offline
Alexey, did you read my explanation(s)?

While our solutions are the same, I'm saying the the main change in peak level is caused by the relative phase changes of minimum phase processing and only partly by the filtering ringing while you're saying it's all caused by the filter ringing and that filter ringing is lower in linear phase.

That doesn't seem right to me.
Alexey Lukin
Verified Member
#21
19th April 2012
Old 19th April 2012
  #21
Lives for gear
 
Alexey Lukin's Avatar
 
Joined: Dec 2007
Posts: 959

Verified Member
Alexey Lukin is offline
Yes, you make a good point that linear-phase filters cause less problems with peak levels, esp. when only small parts of the signal are being cut. However you attribute this to phase relationships, which is not the direct cause of peak levels increase. Even if the filter is perfectly linear-phase (i.e. does not change phase relationships in the signal), there will be increase in peak signal levels dependent on the amount of ringing in the filter's impulse response. For linear-phase filters, this amount of ringing is dependent on steepness of the frequency response change and amount of frequencies cut/boosted. You can see this effect by filtering a square wave with a linear-phase filter.
Lagerfeldt
Verified Member
#22
19th April 2012
Old 19th April 2012
  #22
Audio Alchemist
 
Lagerfeldt's Avatar
 
Joined: Sep 2004
Location: Copenhagen, Denmark
Posts: 5,157

Verified Member
Lagerfeldt is offline
Quote:
Originally Posted by Alexey Lukin View Post
Yes, you make a good point that linear-phase filters cause less problems with peak levels, esp. when only small parts of the signal are being cut. However you attribute this to phase relationships, which is not the direct cause of peak levels increase. Even if the filter is perfectly linear-phase (i.e. does not change phase relationships in the signal), there will be increase in peak signal levels dependent on the amount of ringing in the filter's impulse response. For linear-phase filters, this amount of ringing is dependent on steepness of the frequency response change and amount of frequencies cut/boosted. You can see this effect by filtering a square wave with a linear-phase filter.
Yes, I realize that there will still be a small increase in the peak level even with linear phase (which I also write above in my explanation) and it's easily verifiable in practical tests.

But I'm still arguing that the large(st) change is caused by the phase changes in minimum phase processing and not simply the ringing, i.e. the relative shift in phase causes the frequencies to interact in a way that may or may not lead to a peak change in the new waveform where the troughs and crests correlate differently. Since this is offset during linear phase processing it does not happen, but there will still be some ringing.

Quote:
For linear-phase filters, this amount of ringing is dependent on steepness of the frequency response
But this is also true of minimum phase filters, i.e. the steeper the slope, the higher the ringing.
Alexey Lukin
Verified Member
#23
19th April 2012
Old 19th April 2012
  #23
Lives for gear
 
Alexey Lukin's Avatar
 
Joined: Dec 2007
Posts: 959

Verified Member
Alexey Lukin is offline
Quote:
Originally Posted by Lagerfeldt View Post
But I'm still arguing that the large(st) change is caused by the phase changes in minimum phase processing and not simply the ringing
Phase shifts are connected with ringing, but not directly. Here's a counter-example. The square wave has been filtered by a gentle minimum-phase LP filter:



You can see that there is no ringing and no level overshoot. Here is the frequency response of the filter:



On the other hand, if you apply a similar (yet somewhat steeper) linear-phase filter, ringing and level overshoot will immediately occur:



This is not a typical example, but it shows that the change in peak levels is produced directly by filter ringing, which, in turn, totally depends on the filter's frequency and phase response.
#24
19th April 2012
Old 19th April 2012
  #24
Lives for gear
 
stinkyfingers's Avatar
 
Joined: Apr 2011
Location: Vermont
Posts: 2,981
My Recordings/Credits

stinkyfingers is online now
hmmm...now i am very confused because i am getting some conflicting results with similar "tests"...
i need to get my sh!t together and try and explain myself somehow...:P

edit: ok...i see you have minimum phase showing no ringing...?
i thought there was post ringing in minimum phase...?
hence running it forwards and backwards give you pre & post = linear phase...?
so the actual "level" of the minimum phase ringing would be 2x the linear phase ringing because it's /2 to go both ways for linear...
make sense...?
what am i missing...?
is that ^ post labeled wrong...?
thanks...
__________________
my dog can smell my farts before they happen...
Nordenstam
Verified Member
#25
19th April 2012
Old 19th April 2012
  #25
Lives for gear
 
Nordenstam's Avatar
 
Joined: Aug 2005
Location: Norway
Posts: 1,750

Verified Member
Nordenstam is offline
Hello!

Quote:
Originally Posted by Lagerfeldt View Post
But I'm still arguing that the large(st) change is caused by the phase changes in minimum phase processing and not simply the ringing, i.e. the relative shift in phase causes the frequencies to interact in a way that may or may not lead to a peak change in the new waveform where the troughs and crests correlate differently.
Quote:
Originally Posted by Alexey Lukin View Post
This is not a typical example, but it shows that the change in peak levels is produced directly by filter ringing, which, in turn, totally depends on the filter's frequency and phase response.
I guess this can be seen to depend on the explanation model.

In time domain, there have to be some ringing(in the widest sense). If not, the impulse response is perfect and as such can not do any equalizing.

In frequency domain, it is the phase relationship that contributes to this.

The two models are different viewpoints on the same data.


I've made a bunch of graphs over the years and this is one of my favourites:


Equalizers change the phase response of certain frequencies compared to others. In the lower graph above, the fundamental sine wave in the square wave is offset by 90'. The superposition of the changed fundamental and the unaltered harmonics explains the increase in the peak level. (more here)

Graph made using Izotope RX and mspaint. Check out Falstads Fourier series applet to play with these ideas. Switch to magnitude/phase view to move the different harmonics around.


Cheers,
__________________
Andreas Nordenstam - mastering - music/fractals
#26
19th April 2012
Old 19th April 2012
  #26
Lives for gear
 
stinkyfingers's Avatar
 
Joined: Apr 2011
Location: Vermont
Posts: 2,981
My Recordings/Credits

stinkyfingers is online now
^ahhh...thanks for that...very easy to understand now...
Lagerfeldt
Verified Member
#27
20th April 2012
Old 20th April 2012
  #27
Audio Alchemist
 
Lagerfeldt's Avatar
 
Joined: Sep 2004
Location: Copenhagen, Denmark
Posts: 5,157

Verified Member
Lagerfeldt is offline
Quote:
Originally Posted by Nordenstam View Post

In frequency domain, it is the phase relationship that contributes to this.
...
I've made a bunch of graphs over the years and this is one of my favourites:


Equalizers change the phase response of certain frequencies compared to others. In the lower graph above, the fundamental sine wave in the square wave is offset by 90'. The superposition of the changed fundamental and the unaltered harmonics explains the increase in the peak level. (more here)

Graph made using Izotope RX and mspaint. Check out Falstads Fourier series applet to play with these ideas. Switch to magnitude/phase view to move the different harmonics around.
Thanks for yet another great graph, Andreas.

I think this shows my point exactly, that the relative phase shift between frequencies (can) cause the resultant waveforms to interfact in a way that causes peaks where they were none earlier.

I still fail to see how this could be seen as merely a question of filter ringing since it's precisely the offset of these phase shifts that enable linear phase processing to not exihibit this behaviour - apart from any change caused by filter ringing. To sum up: filter ringing will happen with both minimum and linear phase processing, but the largest change to the peak level will be due to the relative phase shifting in minimum phase processing.

A practical example: a heavily processed and limited kick drum, much like your square wave above, could peak +2 dB when exposed to a minimum phase 12 dB/Oct HPF at 30 Hz. When the same kick drum is exposed to an identical but linear phase HPF it will not peak at all or only a few fractions of a dB, the latter caused by the filter ringing, which switching to linear phase will not get rid off. The main difference lies in the relative phase shifting or absence thereof. Changing the steepness of the filter in any of the two modes of processing will change the peak level more or less in both files, but it's not the direct cause of the main change in peak level.
#28
20th April 2012
Old 20th April 2012
  #28
Gear nut
 
Cubic Spline's Avatar
 
Joined: Nov 2011
Location: France
Posts: 96

Send a message via Skype™ to Cubic Spline
Cubic Spline is offline
If you do an incomplete additive synthesis of a square wave, stopping after 20 harmonics for example, you just create a perfect brickwall linear phase filter with no ringing, and still, the peak level will be superior to the full square wave peak level.

Now shift the phase of one or more of the harmonics (by changing the oscillator init phase), and you do a perfect allpass filter with no ringing, but peak level will still change !

What does it show ? that peak level is more than just filter ringing or phase.
#29
20th April 2012
Old 20th April 2012
  #29
Gear addict
 
Joined: Oct 2011
Location: Los Angeles, CA
Posts: 422

Nonlinear is offline
Quote:
Originally Posted by Trakworx View Post
Try bypassing the Waves SSL compressor and see what happens.
x2

While the discussion about filter ringing, etc., is correct and certainly possible my own experience sides with Trakworx's suggestion. The subsonics are likely pulling down your overall mix level in the compressor which is then feeding the limiter.

If that's the case then you probably have a lot of subsonic energy and would probably benefit overall from some reduction there.

Just my 2 cents!
#30
21st April 2012
Old 21st April 2012
  #30
Gear nut
 
ksound's Avatar
 
Joined: Nov 2010
Location: Phuket
Posts: 124

ksound is offline
Quote:
Originally Posted by Quint View Post
I agree with this. It's possible that the way you have your SSL setup was causing it to clamp down a lot harder when the louder bass frequencies were present. Once you lowered the bass frequency's volume, the SSL quit clamping down as hard and a higher volume level started to make it's way to the limiter, causing it to clamp down harder to make up for lighter amount of gain reduction that the SSL is now contributing.
Definitely the most likely reason.

I find the Waves SSL comp to be ultra sensitive to those low frequencies. Easy to confirm too, compare the gain reduction on the meter in the plugin with the HPF engaged and bypassed.
New Reply Submit Thread to Facebook Facebook  Submit Thread to Twitter Twitter  Submit Thread to LinkedIn LinkedIn  Submit Thread to Google+ Google+ 
 
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
sneezebomb / So much gear, so little time!
7
BLueROom / So much gear, so little time!
26
Roland / Remote Possibilities in Acoustic Music & Location Recording
2

Forum Jump

SEO by vBSEO ©2011, Crawlability, Inc.