middleaudio.com

Lately I've been tracking and mixing to try to obtain the maximum amount of headroom. The reason I say "lately" is because at first I was looking at it from the perspective that I needed to get a nice strong signal going in first (read: ~-12dbfs) but then I'd run into situations sometimes where I'd be having to watch my headroom more than mix.

Nowadays I try to keep my mix bus around -12dbfs so the tracking itself is usally much lower. When I started I guess I wanted my mixes to sound louder so I'd have mixes in some cases that were -3dbfs on average. This isn't something I've ever really had a discussion with anyone about though so I'm curious about it now. When most of you finish your mixes how hot are they? I, of course, am talking about mixes before mastering.

I'm actually surprised that Andrew and I have never mad a real converation about this before... but now's a good a time as any.

I'm about the same, my mix bus usually averages around -12dBFS. Although for songs that build a lot, usually by the end the average will be around -9. Peaks never more than -3. If I start seeing peaks around -3 I'll pull the master fader back. Once in a while, particularly when there's analogue gear on the 2bus, I'll end with peaks close to zero, which I can live with, but I watch it very carefully. I always shoot for tracking around -16, but sometimes creeps up to -12. The nice thing about gain structure when summing digitally, at least in pro tools, is that the master fader is pre bit reduction from 48 to 24, so bringing the fader doesn't do any damage.

Another note, for me drums and percussion often end up a bit hotter for me because the analog gear on the way in will be showing around 0VU, but due to their transient nature end up hotter on the digital peak meters.

Do you ever use a meter with finer resolution for your mix bus? I often wonder about how accurate the meter is in Pro Tools.

No because I don't care how fine it is, because I should never be near a point where it's an issue. I'll use a few different meters, but I'm always looking for things other than just pure level. When mastering I have a whole raft of meters, and I more or less ignore the PT meters.

An engineer just sent me some guitar tracks for a client I'm doing vocals for this weekend. Peak? 0dbfs. Average: -3dbfs

And the worst part is that its probably already been significantly clipped, but the clips aren't showing up any more and there's nothing you can do about it.

macrae11 wrote:And the worst part is that its probably already been significantly clipped, but the clips aren't showing up any more and there's nothing you can do about it.

Yep. Whatever... I am being hired to mix, and mix it I will.

I have received many projects where the first thing I did was trim all the tracks down 6dB before I it play. What was the line...? Yellow is the new Red?

So after lurking on this thread, I went out to teh webz to do do a lot of research. I now understand why peaks around 0dbfs is stupid. But I have a question, still:

1) In any DAW, are levels automatically expressed in dbfs? Or does it vary from program to program? I use Cubase VST 32, so obviously my more specific question is if my DAW expresses meters in dbfs :)

In the past, I've always tracked instruments as close to 0 in my DAW as possible (normalizing when necessary to get the volume up), not because I was consciously in a loudness war, but because I just wanted clear signals. For example, if I recorded a microphone set too low, my vocals sounded real hissy when I made them louder in the program, so I just tried recording with clarity of signal as my goal.

Any digital peak meter will measure in dBFS, so yes Cubase measures in dBFS. One easy way to tell is that dBFS is almost the only dB scale where the highest point is 0.

Christian LeBlanc wrote:In the past, I've always tracked instruments as close to 0 in my DAW as possible (normalizing when necessary to get the volume up), not because I was consciously in a loudness war, but because I just wanted clear signals. For example, if I recorded a microphone set too low, my vocals sounded real hissy when I made them louder in the program, so I just tried recording with clarity of signal as my goal.

This is a gain structure issue. It's possible you were recording too low, and then when you bring up the signal level, you would also be bringing up the inherent noise recorded with that signal. Or you could have been recording it at the correct level and then be bringing it up way too hot, but that should only be an increase of 12-16dBFS, which shouldn't be a huge problem, unless your preamp/mic/ADC signal chain is excessively noisy.

macrae11 wrote:Any digital peak meter will measure in dBFS, so yes Cubase measures in dBFS. One easy way to tell is that dBFS is almost the only dB scale where the highest point is 0.

I was confused for a bit when I looked in Cubase, as it actually goes up to 6.0. Per the instructions: "The VST Channel Mixer allows you to boost weak signals by +6 dB, if you like. Just be sure to avoid signal levels above 0 dB (clipping)." When I set my volume faders up to 6, peaks (on a hot, normalized track) reach around +4.2. So as not to confuse myself, I think I should just pretend those >0 numbers don't exist, and just aim to track around -12 to -16.

Sorry, need to clarify. The numbers you're seeing on your Cubase faders are actually the fader level, not the audio level. So for example if you had audio recorded at -16dBFS and you moved the fader to the maximum value your audio would increase in volume to -10dBFS. The fader values are just the relative increase/decrease by moving the fader. Haven't seen your version of Cubase(or any version) in forever, so I forgot what the mixer looked like.

If you look at this image of the Pro Tools mixer you can see the fader level on the left and the audio level on the right of the meters.
http://www.4creator.com/blog_img/62//protools_mixer.png

After looking at a couple of old Cubase pictures, I believe the number at the top is the highest peak detected on the track, and the number at the bottom is the fader position. Don't see any hash marks to indicate level in real time. Maybe Matt can clarify that.

How do the rest of y'alls DAW meters handle this? I've always heard so many complaints about PT meters, but they've never bothered me, and they seem superior to Cubase's way of handling it visually. Of course this doesn't speak to their accuracy...

Thanks for the clarification! I can see where I got confused, since, you're right, the number at the bottom is the fader, while I only see audio level numbers for clipping (I had a 0.0 at the bottom for my volume faders, but a 7.9 for where it peaked) (which I'd show you, if I had a flickr account)

To further clarify Andrew's point: When your fader level is at 0 it's considered unity gain. Meaning you are not increasing or decreasing the gain.

In Cubase you will see two numbers displayed below the meter. The larger number at the bottom represents the fader level, and the smaller number just above it represents the actual level of audio passing in DBFS. Easiest way for you to see this is by playing back any of your projects. As the audio passing through you'll see the audio level fluctuate in real time whereas the fader level stays static since you aren't moving the faders.

Interestingly enough though Cubase will show levels in DBFS above 0 which I always found really strange... Might be because it uses 32 bit floating point. That being said if you have anything above 0dbfs in Cubase when you export to fixed point 24 bit, don't expect anything above that 0dbfs to not be clipped.

Mathieu Benoit wrote:Interestingly enough though Cubase will show levels in DBFS above 0 which I always found really strange... Might be because it uses 32 bit floating point. That being said if you have anything above 0dbfs in Cubase when you export to fixed point 24 bit, don't expect anything above that 0dbfs to not be clipped.

It is precisely the floating point math that allows you to produce levels above 0dBFS in the DAW. You need to have a solid understanding about digital levels to deal with everything in that domain, especially when dealing with the various ways that different systems crunch the numbers. Levels above "Zero" need to be considered out of bounds, and to be dealt with accordingly.

Mathieu Benoit wrote:To further clarify Andrew's point: When your fader level is at 0 it's considered unity gain. Meaning you are not increasing or decreasing the gain.

This is depending on how you have your pan law set up in the DAW. Some faders at "zero" may mean a level below unity.

Malcolm Boyce wrote:

Mathieu Benoit wrote:Interestingly enough though Cubase will show levels in DBFS above 0 which I always found really strange... Might be because it uses 32 bit floating point. That being said if you have anything above 0dbfs in Cubase when you export to fixed point 24 bit, don't expect anything above that 0dbfs to not be clipped.

It is precisely the floating point math that allows you to produce levels above 0dBFS in the DAW. You need to have a solid understanding about digital levels to deal with everything in that domain, especially when dealing with the various ways that different systems crunch the numbers. Levels above "Zero" need to be considered out of bounds, and to be dealt with accordingly.

I understand all of that, but I find it strange that Cubase actually shows the levels about 0 dbfs on the meter. In Pro Tools it doesn't actually show the level going above 0 dbfs although it too is 32 bit floating point. Does Sonor show levels above 0 dbfs? I just don't want to confuse Christian into thinking that anything above 0 dbfs exists in a tangible sense.

Malcolm Boyce wrote:

Mathieu Benoit wrote:To further clarify Andrew's point: When your fader level is at 0 it's considered unity gain. Meaning you are not increasing or decreasing the gain.

This is depending on how you have your pan law set up in the DAW. Some faders at "zero" may mean a level below unity.

RIght. That's an excellent point, I forgot all about that. That's actually a good discussion to have...

Mathieu Benoit wrote:

Malcolm Boyce wrote:

Mathieu Benoit wrote:To further clarify Andrew's point: When your fader level is at 0 it's considered unity gain. Meaning you are not increasing or decreasing the gain.

This is depending on how you have your pan law set up in the DAW. Some faders at "zero" may mean a level below unity.

RIght. That's an excellent point, I forgot all about that. That's actually a good discussion to have...

I actually tried looking this up in my documentation, but I didn't even find a reference to dbFS, much less how to recalibrate the faders with respect to different places on different db scales, which I understand some DAWs allow you to do.

In my Cubase (and even my Cubase LE 4), it still doesn't show actual audio levels in real time, just peaks. In that sense it's real time, if a track gets louder and louder. Stupid thing, it's going to force me to use my ears instead of my eyes - that's not what I signed up for!

It's not just about your ears though, I think it's a very good thing to have a solid understanding of the theory behind the sound. Those numbers do mean something after all.

Christian LeBlanc wrote:I actually tried looking this up in my documentation, but I didn't even find a reference to dbFS, much less how to recalibrate the faders with respect to different places on different db scales, which I understand some DAWs allow you to do.

In my Cubase (and even my Cubase LE 4), it still doesn't show actual audio levels in real time, just peaks. In that sense it's real time, if a track gets louder and louder. Stupid thing, it's going to force me to use my ears instead of my eyes - that's not what I signed up for!

As far as I know, Cubase won't let you "calibrate" your levels, however you can calibrate the colors associated to them on the meters. For example you can set Yellow to -12dfs and Red to -3dbfs to give yourself some headroom to work with. It's the audio equivalent of setting your watch 5 minutes fast so that you are always early.

For stereo pan law Cubase does let you select your preference. If you go in the Project Setup window (shift+s) you can select from 3 different options IIRC:

0db which means there is no attenuation in the center. Basically anything panned in the will sound louder than it should.

-6dB which means that there is 6dB of attenuation in the center. Basically everything will balance out very well for mono but if you want some extra kick in the center you'll come up short.

-3dB This is basically a happy medium between the two extremes. I'm pretty sure Cubase defaults to this.

Mathieu Benoit wrote:

Malcolm Boyce wrote:

Mathieu Benoit wrote:Interestingly enough though Cubase will show levels in DBFS above 0 which I always found really strange... Might be because it uses 32 bit floating point. That being said if you have anything above 0dbfs in Cubase when you export to fixed point 24 bit, don't expect anything above that 0dbfs to not be clipped.

It is precisely the floating point math that allows you to produce levels above 0dBFS in the DAW. You need to have a solid understanding about digital levels to deal with everything in that domain, especially when dealing with the various ways that different systems crunch the numbers. Levels above "Zero" need to be considered out of bounds, and to be dealt with accordingly.

I understand all of that, but I find it strange that Cubase actually shows the levels about 0 dbfs on the meter. In Pro Tools it doesn't actually show the level going above 0 dbfs although it too is 32 bit floating point. Does Sonor show levels above 0 dbfs? I just don't want to confuse Christian into thinking that anything above 0 dbfs exists in a tangible sense.

Sonar does show levels above zero as well. It is handy, and I wouldn't love not having it.

As Andrew already mentioned, I remember being less than thrilled with the metering in ProTools compared to Sonar.

Malcolm Boyce wrote:Sonar does show levels above zero as well. It is handy, and I wouldn't love not having it.

It's really a non-issue for me since I don't make it a habit of having levels anywhere near there.

Mathieu Benoit wrote:0db which means there is no attenuation in the center. Basically anything panned in the will sound louder than it should.

-6dB which means that there is 6dB of attenuation in the center. Basically everything will balance out very well for mono but if you want some extra kick in the center you'll come up short.

-3dB This is basically a happy medium between the two extremes. I'm pretty sure Cubase defaults to this.

Panning taper or "pan law" is designed to keep the level consistent regardless of where the pan pot is set. If all a pan did was reduce the level of one side while panning to the other, you would be lowering the volume of a channel by panning it off. As you pan off, you are reducing the one side, while increasing the other side to maintain an overall consistent level.

+3dB hard panned is the constant power setting and is that way to make up for the dropping of the additional source, which is the other channel that you attenuated when you panned away from it. If you basically muted the "left half" of a channel, you would see a 3dB drop in level.

+6dB hard pan is a whole other kettle of fish.

The "minus" at center thing in DAWs and other digital things is a clever addition that they've given us. You basically have your choice of being at unity at center, and boosting above unity as you pan off, or being 3dB down at center, and at unity hard panned.