Master Clipping

[Trum]

Hi all,

I have a quick question about clipping and loudness. I'm working on a track at the moment (probably only my third in Ableton? I only just got it and I love it!), and I've got about about seven channels that are all nice and healthy, but the master is right up in the red. My problem with that is that nothing is distorting, and I really love the overall mix. I can bring it down out of clip but the track sounds flat when I do.

I guess this leaves me with a couple of questions.

i) How significant is clipping at the master if nothing is distorting, and the volume of the track is roughly equivalent to commercial releases? I doubt this will ever get played through a big enough system to worry about it blowing anything up, and I'd imagine a big system would at least have a limiter or something to prevent that from happening anyway.

ii) At the pre final-mix stage, how important is volume? Should I be looking for a good "mix" that is quiet, and then try to compress/EQ/boost it once it has been mixed down?

These are problems I've always encountered but never really thought to address. Ironically, most things I've produced are quieter than I would like but I have a habit of running loud during creation. Would it be wise to move volume concerns to later in my workflow? I really enjoy working with headphones and having it pretty loud. I do have a headphone amplifier I could use to negate that point and keep everything out of clip. That clip meter must be there for a reason, even if it's just there to be a buzz kill or whatever.

I'm guessing I don't need to re-invent the wheel here, and there is a standard way to approach this aspect of song creation. I would love to hear your thoughts.

Thanks!

[Shift Gorden]

Good question! Yeah, you can clip within Ableton without distortion...but once you render out of Ableton you'll be in a whole lot of trouble

Here's an experiment for you since you said turning down the master results in a flatter sound...when you turn down the master, turn up the volume on your computer/soundcard - I bet it's not flat sounding any more. Louder stuff sounds better - science.

As you get more and more into Ableton, you'll probably start to focus on your mixes and masters more. Instead of turning down the master, keep that at 0db. Instead, focus on balancing out your 7 channels - turn those guys down instead. Ideally, you'll want about -3db peak on your master track. That way, you can add some bits and pieces to your master if you need to (like you mentioned, EQ, compression, saturation, etc).

At mix-down, volume is super important. Again, I wouldn't worry too much about things being too quiet. I'd try to focus on introducing balance into your workflow (and crank up your headphone amp since that sounds pretty important to you when you're creating).

Just a few ideas from a fellow hobbyist.

[Trum]

Thanks for the reply!

I actually run my headphone / computer volume on 100% while working, so the only room I have to move there is with the headphone amp, haha. That's what I meant about it sounding flat. When I bring the track down out of clip using the channels, it just sounds flat to me compared to commercial releases. I like to volume check against songs that I like the mix of, and so far the only way I ever get close is having the master in the red. I guess this is what lead me to wonder if volume is added in post-production.

[Tarekith]

Typically the volume of commercial tracks is boosted in the mastering stage, which is the very last thing you do when working on a track. Prior to that you definitely are better off not driving the master too hard if you can help it, it's just a good practice to get into so you don't inadvertantly clip. Nothing wrong with cranking the master a little if you want to compare your track against commercial releases once in awhile, but I wouldn't leave it cranked when working on it most of the time. Just turn up the headphone amp instead.

Here's a bit more about mixing and mastering that might help clear up the process for you:

http://innerportalstudio.com/articles/Mixdowns.pdf

http://innerportalstudio.com/articles/Mastering.pdf

[Trum]

Typically the volume of commercial tracks is boosted in the mastering stage, which is the very last thing you do when working on a track. Prior to that you definitely are better off not driving the master too hard if you can help it, it's just a good practice to get into so you don't inadvertantly clip. Nothing wrong with cranking the master a little if you want to compare your track against commercial releases once in awhile, but I wouldn't leave it cranked when working on it most of the time. Just turn up the headphone amp instead.

Here's a bit more about mixing and mastering that might help clear up the process for you:

http://innerportalstudio.com/articles/Mixdowns.pdf

http://innerportalstudio.com/articles/Mastering.pdf

Thanks for the links! Will absolutely have a read.

[Stromkraft]

When I bring the track down out of clip using the channels, it just sounds flat to me compared to commercial releases.

When you imagine things sound different regarding to volume it's a good idea to investigate by isolating the volume differences so that you stop being biased by volume preference, ie louder-is-better syndrome.

One quick step is to lower your master so it's not in the red and increase the volume of your audio interface by exactly the same amount, as suggested by Tarekith. If this sound equally good chances are it was only the volume difference that led you to believe going into the red sounded "better". If not, the answer is likely found post the Live output.

Phase Cancellation as a tool
Sooner or later as a music maker you will encounter the phenomenon of "phase cancellation", ie the fact that the same audio signal, no matter how complex, playing back in time with an identical signal, at the same volume and with the polarity switched (or phase switched) will cancel out the resulting signal to silence or near silence (depending on the degree of match).

This phenomenon can also be used as a tool to evaluate for example whether two signals are the same or not. For example this is used by Ableton for several of their own tests for the Live audio engine as described in parts of the Live 9 manual.

The basic, well-founded conclusion to understand here is scientific and concerns the fact that if 2 signals, one of them having its polarity, or phase, switched, cancels out then these tested signals are exactly the same. What you think you hear therefore must be wrong or at least not be found in the Live mixer.

Test your assumptions
Here's a way to find out if you're just imagining things with regard to what levels you use in your tracks:

• Run all your tracks now going to Master to a buss, call it "Loud Buss" or something that identifies this as the original for you. The fader for all busses in this test should be set at 0dB. All busses also need have "no input" set and monitor set to "In".
  
  If you don't see 2 values as in my screen dumps, hold the mouse pointer just on the divider just above the scale of any track so that the mouse pointer changes into a double arrow with a horizontal line trough it, and hold down the mouse and drag upwards.
  …
• Make sure you get the same peak values, in the "red", in this "Loud Buss" as in the master. This may not be as visually indicated as in the master, but the values should be exactly the same.
  
  Here I've put a drum and a bass track that together peaks at +4.17dBFS, which is actually not true visavi the Live audio bus which is 32bits and has at least 190dB of dynamic range** (See comment about this at the end).
  
  
  
  Note routings and fader levels.
  …
• Route "Loud Buss" to another audio track that you can call "Loud Compare". This buss is routed to "Master", which is now just a channel going out to your audio interface. Why to do this is explained later.
  
  See how I did it:
  
  
  …
• Now select all the tracks routed to "Loud Buss" and the buss itself and choose duplicate (Right-click for contextual menu and choose or go to the Edit menu.
• Rename the duplicated buss something like "Lower Master".
• Change the routing of "Lower Master" directly to Master and solo it. Also make sure all duplicated tracks are playing and that you have the same volume meter readings in this duplicated buss as in the original.
  
  
  
  The duplicated tracks and modified new buss.
  …
• Add the native audio effect Utility to "Lower Master" and change the gain of this effect to -6dB. If you need to go lower than that in order to not go into the red use this number instead of -6 everywhere in this description. The meter readings should now be below the red area. You may need to click once in the peak values square to make it reset.
• If you solo your 2 last busses — one at a time of "Loud Master" and "Loud Compare"— you should now hear the same thing only at different volume levels. Which do you prefer?
• Now, let's even the playing field by adding Utility also to the "Loud Compare" Buss. Reduce the gain of Utility to -6dB as in your "Lower Master" buss and also check the left and right phase buttons below:
  
  
  
  This setting will reverse the polarity, or the phase, of the track. The ability do this on one signal path and not the other is why we needed this buss step on the way to the master output.
  Again solo the busses on at a time to make sure it sounds as expected.
  …
• Here is the final verification step. Making sure you are playing back all the clips of the same scene or are completely playing back only tracks from Arrangement, turn of all solo buttons.
  
  At this point, with all tracks playing, you should see the meters working but only silence should be heard.
  
  What does this mean?

Conlusions

Well, if there is silence both busses have identical signals (before the reversed polarity). They are the same and the only difference is volume.

If sound still can be heard the signals are not identical. Review all steps because they should be. If they are you need to find out how they are different.


**It's more useful to think of this overshoot as relative to Full Scale in 16 bits. AFAICT Live displays all bit depths the same, yet because the internal signal path are 32bit audio you can go over 0dBFS without a direct penalty.
0dB is the maximum possible digital audio level and if you truly allow your signal to go above 0dB you will clip, period.

[jlgrimes]

Hi all,

I have a quick question about clipping and loudness. I'm working on a track at the moment (probably only my third in Ableton? I only just got it and I love it!), and I've got about about seven channels that are all nice and healthy, but the master is right up in the red. My problem with that is that nothing is distorting, and I really love the overall mix. I can bring it down out of clip but the track sounds flat when I do.

I guess this leaves me with a couple of questions.

i) How significant is clipping at the master if nothing is distorting, and the volume of the track is roughly equivalent to commercial releases? I doubt this will ever get played through a big enough system to worry about it blowing anything up, and I'd imagine a big system would at least have a limiter or something to prevent that from happening anyway.

ii) At the pre final-mix stage, how important is volume? Should I be looking for a good "mix" that is quiet, and then try to compress/EQ/boost it once it has been mixed down?

These are problems I've always encountered but never really thought to address. Ironically, most things I've produced are quieter than I would like but I have a habit of running loud during creation. Would it be wise to move volume concerns to later in my workflow? I really enjoy working with headphones and having it pretty loud. I do have a headphone amplifier I could use to negate that point and keep everything out of clip. That clip meter must be there for a reason, even if it's just there to be a buzz kill or whatever.

I'm guessing I don't need to re-invent the wheel here, and there is a standard way to approach this aspect of song creation. I would love to hear your thoughts.

Thanks!

i) This is ultimately up to you to decide. 1 to 2 db of digital clipping might not be audible or desirable in some cases (especially if the clipping is brief for some stray peaks). If the clipping is for an extended period, you probably will hear the clipping as your audio basically flatlines when it clips. Just keep in mind if you are taking this file to be mastered, you will probably be making their job harder as they prefer to do the extreme limiting on their end as they usually have better tools/experience in extreme limiting (if it is even needed) and extreme limiting is something that is basically undoable if you went too far. Welcome to the loudness wars! There is alot of debate in this, but IMO if you are heavy limiting a track and it is going to another limiter on an external system, your audio quality will probably get worse. While most people agree that you shouldn't leave any mixing decisions to mastering engineer (as their job is to mainly do small touch up adjustments and provide a second ear before final product goes out), limiting is about the only exception and most folks recommend leaving limiting of master bus to master engineer.


ii) There is some debate to this, but what I hear most is don't worry about getting your tracks as loud as commercial releases at the mixdown stage. Most people recommend having your mixdown peak around -6db to -3db of full scale meter. That said I rarely pay attention of where my master fader is peaking as it is often well below clipping (usually somewhere from -12db to -6db). It sounds like you are "mixing in the red", while I'm sure you are getting a good sounding mix in the red (it happened to me before as well), you are not hearing your mix properly and are possibly making some bad judgements doing so. Thats probably why when you turn down your master fader, your mix comes apart. I don't think most masterning engineers mind a little bit of compression /EQ on the master bus as if done correctly can help "gel" the mix. They tend to hate heavy limiting on master bus. (Notice I said master bus, if a limiter helps fatten your individual tracks, or busses then by all means do it, just be mindful that you can get so far where the mastering engineer won't be able to help if track as no dynamics). What I tend to do is mix with a lot of headroom with no digital clipping and when I feel the mix is about complete, I slap a limiter on the master bus and crank it up to get an idea of how it will sound mastered, from that point I will make some adjustments to get the "pseudo-mastered" mix sound better and then remove the limter plug-in from there and export mix.



All in all if you dont plan on mastering song I wouldn't worry too much about it though as you said you like the mix. I would though consider for future projects not going in the red on the master bus and learn how to get a fat sounding mix without master bus clipping (especially if you plan on getting your songs mastered at some point).

[Trum]

When I bring the track down out of clip using the channels, it just sounds flat to me compared to commercial releases.

When you imagine things sound different regarding to volume it's a good idea to investigate by isolating the volume differences so that you stop being biased by volume preference, ie louder-is-better syndrome.

One quick step is to lower your master so it's not in the red and increase the volume of your audio interface by exactly the same amount, as suggested by Tarekith. If this sound equally good chances are it was only the volume difference that led you to believe going into the red sounded "better". If not, the answer is likely found post the Live output.

Phase Cancellation as a tool
Sooner or later as a music maker you will encounter the phenomenon of "phase cancellation", ie the fact that the same audio signal, no matter how complex, playing back in time with an identical signal, at the same volume and with the polarity switched (or phase switched) will cancel out the resulting signal to silence or near silence (depending on the degree of match).

This phenomenon can also be used as a tool to evaluate for example whether two signals are the same or not. For example this is used by Ableton for several of their own tests for the Live audio engine as described in parts of the Live 9 manual.

The basic, well-founded conclusion to understand here is scientific and concerns the fact that if 2 signals, one of them having its polarity, or phase, switched, cancels out then these tested signals are exactly the same. What you think you hear therefore must be wrong or at least not be found in the Live mixer.

Test your assumptions
Here's a way to find out if you're just imagining things with regard to what levels you use in your tracks:

• Run all your tracks now going to Master to a buss, call it "Loud Buss" or something that identifies this as the original for you. The fader for all busses in this test should be set at 0dB. All busses also need have "no input" set and monitor set to "In".
  
  If you don't see 2 values as in my screen dumps, hold the mouse pointer just on the divider just above the scale of any track so that the mouse pointer changes into a double arrow with a horizontal line trough it, and hold down the mouse and drag upwards.
  …
• Make sure you get the same peak values, in the "red", in this "Loud Buss" as in the master. This may not be as visually indicated as in the master, but the values should be exactly the same.
  
  Here I've put a drum and a bass track that together peaks at +4.17dBFS, which is actually not true visavi the Live audio bus which is 32bits and has at least 190dB of dynamic range** (See comment about this at the end).
  
  
  
  Note routings and fader levels.
  …
• Route "Loud Buss" to another audio track that you can call "Loud Compare". This buss is routed to "Master", which is now just a channel going out to your audio interface. Why to do this is explained later.
  
  See how I did it:
  
  
  …
• Now select all the tracks routed to "Loud Buss" and the buss itself and choose duplicate (Right-click for contextual menu and choose or go to the Edit menu.
• Rename the duplicated buss something like "Lower Master".
• Change the routing of "Lower Master" directly to Master and solo it. Also make sure all duplicated tracks are playing and that you have the same volume meter readings in this duplicated buss as in the original.
  
  
  
  The duplicated tracks and modified new buss.
  …
• Add the native audio effect Utility to "Lower Master" and change the gain of this effect to -6dB. If you need to go lower than that in order to not go into the red use this number instead of -6 everywhere in this description. The meter readings should now be below the red area. You may need to click once in the peak values square to make it reset.
• If you solo your 2 last busses — one at a time of "Loud Master" and "Loud Compare"— you should now hear the same thing only at different volume levels. Which do you prefer?
• Now, let's even the playing field by adding Utility also to the "Loud Compare" Buss. Reduce the gain of Utility to -6dB as in your "Lower Master" buss and also check the left and right phase buttons below:
  
  
  
  This setting will reverse the polarity, or the phase, of the track. The ability do this on one signal path and not the other is why we needed this buss step on the way to the master output.
  Again solo the busses on at a time to make sure it sounds as expected.
  …
• Here is the final verification step. Making sure you are playing back all the clips of the same scene or are completely playing back only tracks from Arrangement, turn of all solo buttons.
  
  At this point, with all tracks playing, you should see the meters working but only silence should be heard.
  
  What does this mean?

Conlusions

Well, if there is silence both busses have identical signals (before the reversed polarity). They are the same and the only difference is volume.

If sound still can be heard the signals are not identical. Review all steps because they should be. If they are you need to find out how they are different.


**It's more useful to think of this overshoot as relative to Full Scale in 16 bits. AFAICT Live displays all bit depths the same, yet because the internal signal path are 32bit audio you can go over 0dBFS without a direct penalty.
0dB is the maximum possible digital audio level and if you truly allow your signal to go above 0dB you will clip, period.

Dude this is amazing. Thank you for taking the time to explain all this (with screenshots!), I'll definitely be looking into this when I get home.

[yur2die4]

Just curious, as a double-check, which audio drivers do you have selected in Live audio preferences?

Also, not sure which OS you're using, but if you happen to be using Windows, there might be settings turned on that do things to your audio between when it exits Live and when it goes out to your headphones.

These kinds of things are something to take in consideration about how your audio will be played back on other devices, but it's also something to consider where your audio can be manipulated in a surprising way that you may not have taken into account.

This also applies to popular commercial music. Those tracks are often quite resilient when these kinds of processing are applied to them, and they might sound way different when they have those effects applied to them than when you might hear them in their most direct, raw form.

This might not apply at all to your case, but if it does it is definitely something to be aware of.

[fishmonkey]

it's absolutely critical to understand that when making comparisons at any stage, you must level-balance the two tracks first. it's an artifact of our hearing system that a louder track sounds better. this applies when mixing, tweaking effects, comparing tracks, etc.

for example, a classic rookie mistake is to think that an EQ or compression change has improved the sound, simply because the change has made the track a bit louder.

you also mentioned monitoring at high levels. of course it's good to sometimes hear your work nice and loud, however most people will recommend mixing at modest levels. this helps avoid the problem of the loudness tricking you into thinking that your track sounds better than it is, and also gives you more time before your ears get fatigued.

as you have noticed, a well-mixed track will sound good even at quiet levels. and generally speaking, if a track sounds good quiet, it will sound even better when cranked. however, the reverse is not necessarily true...

[Stromkraft]

I've been plugging this, in many cases free, plug-in elsewhere too and 4U+ BlindTest does help in removing volume bias from the equation in a quite elegant way, if signal paths are set up properly.

I should also mention also here that if one adopts gain staging, i e work on maintaining a stable level, IMHO the RMS level rather than the peak, throughout a track in every process stage, it's much easier to determine if the processing is beneficial for the instrument in question.

This does mean that one need to take the care to measure before and after a processing stage at critical times. I feel it's very much worth it. I use FreeG for measuring RMS levels because I've learnt to trust that over Live's meters, which I haven't been able to get a numerical RMS read from. At least I don't know where that can be found.