Clipping Question

[Syncretia]

If I take two channels, stick a kick on each, turn up the volume until they both almost clip, and leave the master fader at 0db, I find that I get clipping on the master track. This is to be expected.

My question: if I turn down the master track is that clipping still occurring? Logically I would think that the clipping occurs at the point of mixing the two signals together and has already clipped when the signals hit the master fader. However, when I turn down the master, I don't see any red, and I can't hear any clipping. How can this be explained?

Is it possible that clipping can occur without seeing the clipping in the chain or next to the faders?

[Tone Deft]

different scenario. each TRACK has about 63 dB of headroom. so, you can raise each kick by about 30dB and still not clip AT THE TRACK LEVEL. the Master out reads true, 0dB is the end of the headroom, signals will clip if they go over 0dB.

[Syncretia]

63 dB of headroom

Seriously? Where did you read that? How is that possible?

signals will clip if they go over 0dB

The mixed signal is definitely above 0db. I know this because the master fader is at 0db, yet the master is going red (clipping).

So there question is: where does that clipping occur? Before the signal reaches the master mixer, or is there some amplification going on after the mixing point even though the master is set to 0db?

[Tone Deft]

an Ableton employee posted that a while back then corrected himself. they wrote 63dB then said it was 62dB, it's in that ballpark.

how can it be? they wrote the software, they can do anything they want.

try some tests yourself to see how the gain staging works. track 1 meter peaking at 10dB, track 2 metering peaking at 10dB, should equal 20dB on the Master if the Master fader is at 0dB.

the clipping occurs at the Master. there's no amplification, the Master has 0 headroom, the tracks have tons. this makes it easier to mix.

[Syncretia]

Well, that might explain a few things, but I'm a little weary of statements like these. Do you happen to have the link of where the Ableton employee posted this?

I'm sorry to question you on this. You might be correct, but I need to be able to confirm the veracity of this statement with empirical evidence.

Anyway, so what your claim is, is this:

-Each track's individual volume is turned down by 62db even though the peak may be displayed at say "0db"
-So when you mix two sounds together they will still be well under 0db before they hit the master mix
-The sounds are then mixed at the master stage. No amplification is done if the master fader is at 0db.
-We only see clipping when the master is pushed so high that the mix goes above 0 db.

Well, on its face, this doesn't seem to make sense. But before I go on, could you please confirm that this is what your claim is? If this not your claim, please outline what you are saying carefully so I can verify or disqualify what you are saying.

The above doesn't really make sense because if something is close to clipping, and then we duplicate it on to a separate track (perhaps with a little bit of a different EQ), we get clipping at the master level. But, going by the claim mentioned above, clipping would not occur because there should theoretically be plenty of headroom to allow for this.

Also, please clear up what you mean by "headroom". I've taked in to mean 62db of negative gain (i.e. they turn the volume down by 62db on each track). If this is not what you mean, please clarify.

how can it be? they wrote the software, they can do anything they want.

Software can't defy the laws of physics though.

PS: I'm asking this question because I want to know whether or not it's possible for something to clip in Ableton even when we don't see any red on any of the volume bars.

[Tone Deft]

I write this crap in my own free time, sorry, but no, I'm not going through all that. it's 8pm, I'm still at work, I'm going home.



do some experiments.

use Operator or audio clips of a single pure sine wave on two tracks, each at different but fairly close frequencies (1k and 2k.)

put Spectrum on the Master. before clipping you will see two independent spikes on Spectrum. when clipping occurs you will see other frequencies start to spike up in higher frequencies.

you can also put Spectrum on each track and see how high you have to raise their volume before clipping occurs.

with that you can do clipping experiments all night long.

good luck, it sounds like you'll enjoy doing this. you sound engaged with the nuts and bolts of Live.

[Syncretia]

Tone. You've made a VERY controversial statement. You may or may not be correct. I'd like to test your hypothesis. But, you haven't yet explained exactly what your hypothesis is. No experiments are going to prove anything without a clear starting hypothesis. Anyway, I have done a quick experiment based on what you said.

-I put a sine wave operator on two separate tracks
-I put the volume on each operator at 0db.
-I set the track volume faders to 0db
-I set the master to 0db
-Played one at C3, and one at C2
-I put spectrums on both tracks and the master.
-When I hit play, both channels peak at exactly -.03db
-The master peaks at 4.88 (clipping its arse off)
-The two spikes on the spectrum are at 130hz, and 260hz (this is true in all 3 spectrums)
-Although the master IS CLEARLY CLIPPING, there are no especially sharp spikes in the higher end.

Anyway, if we take my outline of your hypothesis, it's clearly false because if it were the case that the two tracks had their volume turned down by 62db, the master wouldn't be clipping. However, I could just be attacking a straw man because I don't actually understand your hypothesis completely.

---------------------

But... I have a snippet on information to work with now. If I turn the master down 5db and click on the Master channel, I can see that there clearly should be clipping going on. The first volume meter in the chain located in the far left hand corner of the Ableton screen is going red. This is happening despite the fact that the master is green indicating that no clipping is occurring at that level. But, more importantly, I can't hear any clipping. So, this just leaves me more confused. Ableton's visual display is telling me that something clipped, yet I can't hear any clipping.

Now, I'm even more confused...

[fishmonkey]

your logic is backwards.

the individual tracks are not at -60dB, it's just that their internal clipping point is at +60dB. so, individually you can go above 0dB and not clip an individual track. in contrast, the master has no headroom above 0dB.

the extra headroom on the tracks isn't defying physics, it's possible because of floating point maths.

and by the way, the old thread referred to is here, the +60dB headroom figure is mentioned by NIS near the bottom of the first page:

http://forum.ableton.com/viewtopic.php? ... 1699&hilit

[Syncretia]

Thanks to the both of you! That makes complete sense now! It's also completely destroyed some of my mixing understanding.

I did a test, and I now understand what Tone Deft was saying.

So just to clarify what fish monkey and Tone Deft have been saying.

-The internals of Ableton allow a signal to go up to somewhere around 60db without clipping.
-Some mention was made a "floating point" math, but that's irrelevant. Floating point just means that a number can be represented with different scales and precisions (http://en.wikipedia.org/wiki/Floating_point)
-Just because a signal inside Ableton is going red, it does not mean that clipping has occurred. In fact any track can go well in to the red before it actually clips.
-The most likely place that clipping will occur is at the master track. If the master track actually does go in to the red, clipping HAS occurred.
-When Tone Deft mentioned "headroom", he was saying that Ableton channels will not clip until they reach about 60db (with the exception of the master track).

Wow. This is a real revelation. Just last night I told someone that they should correct some clipping at the start of their chain. Little did I know that it wasn't clipping.

Also, to answer my original question, it is almost completely safe to turn the master volume fader down in order to avoid clipping. Although, I still wouldn't recommend this because if you leave the master at 0db, it's very easy to see how much mastering headroom you have at any point in time.

[fishmonkey]

Thanks to the both of you! That makes complete sense now! It's also completely destroyed some of my mixing understanding.

I did a test, and I now understand what Tone Deft was saying.

So just to clarify what fish monkey and Tone Deft have been saying.

-The internals of Ableton allow a signal to go up to somewhere around 60-64db without clipping.
-Some mention was made a "floating point" math, but that's irrelevant. Floating point just means that a number can be represented with different scales and precisions (http://en.wikipedia.org/wiki/Floating_point)
-Just because a signal inside Ableton is going red, it does not mean that clipping has occurred. In fact any track can go well in to the red before it actually clips.
-The most likely place that clipping will occur is at the master track. If the master track actually does go in to the red, clipping HAS occurred.

Wow. This is a real revelation. Just last night I told someone that they should correct some clipping at the start of their chain. Little did I know that it wasn't clipping.

although there is massive headroom on individual tracks in Live, it's good general practice to mix with a rational gain structure where your master is not clipping when the master fader is set at 0dB.

and the floating point thing is not at all irrelevant. it means that a much larger range of values can be represented with the same number of bits (in contrast to a fixed point number system).

[Syncretia]

I read your mind:

it's good general practice to mix with a rational gain structure where your master is not clipping when the master fader is set at 0dB.

Although, I still wouldn't recommend this because if you leave the master at 0db, it's very easy to see how much mastering headroom you have at any point in time.

[Syncretia]

Actually, I've had further thoughts on this.

It's not that the tracks before the master track have 60db of headroom. My first interpretation of Tone's statement must have correct. The internal channels must all be turned down by 60db, and then after they have all been mixed in to the master track, and the gain has been applied to that from the master fader, the program must throw another 60db of gain on top of that.

The reason I say this is that it's not enough that the internal tracks be able to go higher than 0db. If this were the case, then the signal would still clip at the master track level.

So think that's it. Internally, Ableton mixes everything at -60db, and then throws 60db on at the end.

Quite strange really. I can definitely see why the designers would have done this, but nonetheless it makes for some serious confusion.

[fishmonkey]

nope, you really don't understand how it works. gotta run now, will check back later.

[Syncretia]

Well, please explain to me where I am going wrong.

[fishmonkey]

here's an analogy for you:

think of the individual channels and the master channel as glass jugs.

the individual channels are bigger jugs, with 0dBFS marked on the side and +60dBFS at the top. the master jug is a smaller jug, with 0dBFS marked at the top.

so any of the track jugs can be filled to way over 0dBFS without overflowing. however, when you add contents of the track jugs together and pour them into the master jug, if you go over 0dBFS on the master, it overflows (clips) right away.

btw, another good reason to be conscious about overdriving the track channels is that any plug-ins you are using on a track may (or may not) behave unexpectedly when overdriven.