Headroom - More before or after tracking?

[Phuturistic]

Don't get me wrong - I'm a firm believer in leaving plenty of headroom. My question is centered around recording line level input devices such as sound modules and synths, no mic'd sources. I use Digital Performer 4.6 to sequence MIDI compositions, and record those MIDI tracks to stereo audio tracks (usually I have from 15 to 20 stereo audio tracks I'm working with at mixdown).

For some strange reason, the DP manual says to record as hot as possible when recording to an audio track, but with no clipping in the track at any point (obviously). Makes sense, but at the same time this advice directly conflicts with what most of the smart sound guys/mastering gurus here on this forum recommend. I've found a happy medium, and record most of my MIDI tracks to audio at around -18dbfs to -12dbfs, usually more towards the latter. If were talking about transient-important instruments such as kick and snare, anywhere from -12dbfs to -6dbfs.

I remember reading on this forum somewhere that it doesn't help to record individual tracks at a hotter level while tracking, and then bringing down the levels by a few to several db's once everything is on its own dedicated audio track when the real mixing begins (when I add EQ, compression, gating, whatever is necessary). I just don't understand why this would be. It seems to me that when recording line level instruments, which always have some amount of noise floor, it just seems like it would be wiser to record these signals hotter, but ALWAYS with some headroom, at least 6dbfs of headroom in ANY situation.

I just feel that a stronger signal (and again, only talking about hardware MIDI modules and samplers, synths) would interact better with plug-ins, help with line level noise floor issues, and just result in a better sounding mixdown. And after carving out unnecessary frequencies with EQ sculpting, some carefully applied compression, and gating out noise floors on individual tracks, I'm usually hitting about -8 or -9dbfs on the mix buss (before any "home mastering" is attempted). So, yes, I do believe in headroom.

Any opinions are welcome and thanks for reading.

 

[legionserial]

I have read a lot of manuals that say to track as hot as you can without clipping. But its fairly possible that the guys who wrote the manual aren't the same people as the guys that made whatever software or hardware, and therefore while they may know how it works, they may not be recordists of any kind and may not really know anything about such things.

If you record one track at like -0.1dbfs, and then add another 20 tracks each recorded atat -0.1dbfs, the combination is going to cause clipping on output. So its kinda pointless to track that hot if you are going to have to turn it all down at the master fader anyway.

Also, if you find one of your tracks isn't loud enough in the mix and you want to turn it up, if you have recorded it at -0.1dbfs, then you aren't going to be able to turn it up very far before that track clips. Meaning you will probably have to turn everything else down. Which is just a pain in the arse.

 

[SouthSIDE Glen]

The "hot as possible without distorting" used to be good advice in the days of analog and even in the days of 16-bit recording.

But if you're recording pure MIDI/digital these days you should be doing 24-bit recording. At 24-bit, that "hot as possible" idea is not only no longer necessary, it's also now not the most adventageous idea.

The idea used to be with analog and 16-bit to get as much dynamic range between your peaks and your noise floor as possible. But now at 24-bit, dynamic range is frankly a non-issue; so it's wiser to sacrifice a little of that extra range it gives and save that for downstream headroom instead.

I think all these manuals that say "hot as possible" are just copying from older manuals.

G.

 

[legionserial]

I'm not sure I understand this correctly, but if you are mixing down midi tracks to audio files (I do the same with each drum piece with a midi drum sampler), then surely the noise floor of each of those tracks is going to be pretty damn low anyway. Assuming you are mixing down within the software, as opposed to running it out and back in again.

 

[SouthSIDE Glen]

I'm not sure I understand this correctly, but if you are mixing down midi tracks to audio files (I do the same with each drum piece with a midi drum sampler), then surely the noise floor of each of those tracks is going to be pretty damn low anyway.

Sounds to me like you understand it perfectly correctly .

G.

 

[FALKEN]

dont forget about the headroom on your synths and sound modules.

 

[SonicAlbert]

dont forget about the headroom on your synths and sound modules.

Really good point.

People often forget that when it comes to synths there is headroom to consider even before the signal is turned into sound. That is, midi headroom. If you are using low velocities or high velocities with low continuous controller 7 data, then you are already not getting the best signal to noise ratio out of your synth. In addition, sometimes people turn the master volume on their synths down, sort of mixing it from the front panel. Also a huge mistake, as synths should have the volume turned up to 100% at all times when recording or performing. This gives the best signal to noise ratio.

In other words, if you turn down the volume of the synth from the front panel, and are using low midi velocities and cc 7's, you have to turn up the volume later, which means you are also turning up the noise floor.

So if you start with the synth up 100% and are using the internal tone generator to its fullest, you are already well on the way to getting the best sound out of the synth possible. It means that the synth has to be turned up less later, and that the trim and channel volume on the mixer can be more optimized for the best performance of that piece of gear.

I say this only as a response to the comment that the noise floor on a synth is low. Yes it is, if the midi data and master volume are handled correctly.

 

[Reggie]

I'm not sure I understand this correctly, but if you are mixing down midi tracks to audio files (I do the same with each drum piece with a midi drum sampler), then surely the noise floor of each of those tracks is going to be pretty damn low anyway. Assuming you are mixing down within the software, as opposed to running it out and back in again.

Well, part of the stuff you hear lately about recording peaks at -14 or whatever has to do with giving yourself digital headroom so your tracks aren't clipping when you boost an EQ or something; but another big part of the reasoning is that 0dbFS is like 18db above the optimal level for a lot of analog gear. Pushing your preamps and stuff hot enough to reach 0dbFS can sometimes cause crappier sound as stuff overloads. Of course this part has nothing to do with your situation ITB; just thought you should know.

 

[Phuturistic]

Great points all around.

SonicAlbert, in response to your recommendation that the synth/sound module be turned up all the way on the hardward volume knob, I always do this as habit. Never does it move from 100%. I use a pair of E-MU hardware sound modules, and even with the volume turned up 100%, along with adjusting midi velocities to their optimal levels (usually anywhere from 75 to 127, of course this entirely depends on the tone I'm trying to acheive with the patch, as different midi velocities can change the tone or timbre of each patch drastically), I still often hit levels recording to an audio track as low as -24dbfs, even with CC#7 pushed up to 127, and I'm referring to the peak levels that patch is hitting. Maybe that doesn't sound that low to some of you, but leaving 24db of headroom seems to be excessive to me in this type of situation.

I actually bought a decent pre-amp (a Brent Averill 2 channel 1272) to assist with these situations where my levels seem to be too low. I'm always very careful with the amount of gain I apply to my sound modules using the pre-amp, as I do not want to create excessive noise floor levels. I find that when I remove the 1272 from the chain, my noise floor sits around -75dbfs, and tracks peak hardest (usually snares and kicks) at about -18dbfs, maybe a little lower than that. With the 1272 in the chain, noise floor jumps up to around -65 to -60dbfs, and tracks can be recorded with the hardest peaks hitting around -12dbfs, sometimes as hot as -6dbfs.

Those numbers tell me that the S/N ratio seems to stay pretty consistent, so its just a matter of do I want to record everything a little hotter, with hotter peaks and a hotter noise floor, or leave things at a lower level when tracking and bring them up when rendered to audio. Its probably a moot point, but my instincts tell me to track with peaks around -12dbfs (while using the pre-amp), and pushing the faders down a little when mixing my stereo audio tracks. I think its because I hate to see them (the stereo audio tracks that have already been rendered from MIDI) pushed over 0 vu on DP's metering, and above the fader level of 0 vu on the mix buss.

Legionserial, if I was to record individual tracks at 0.1dbfs each (which I would never do, as I could never achieve that level with many of the patches I use), I would push down the levels of my audio tracks that I recorded my MIDI tracks into, so it would never overload the mix buss (master fader). I'm speaking mainly of the level individual MIDI tracks are being recorded at when going to their own dedicated audio tracks (before adjusting the levels of the audio tracks). And I don't have any problem turning my audio tracks down, knowing that I'm doing so in an effort to prevent overloading the mix buss.

 

[SonicAlbert]

I just checked my Emu Proteus 2000 and Vintage Pro channels on my mixer and it does seem I have them jacked up kind of high. It's weird because I've never felt a shortage of volume when using them live, but I too notice the thing where I want them louder in my studio.

-75 is really a pretty good real world noise floor for an analog studio. Seriously. I'm having a problem now where something is bleeding noise into my system, and it's raising the noise floor badly. I've tracked it down to a group of external power supplies, but I haven't torn apart the studio yet to really troubleshoot it. I'm moving in a few weeks, so I'll fix it when I set up in the new place. But -75 looks really good to me right now.

Of course, when I work ITB the noise floor is way down there, as you'd expect.

I do also use a preamp sometimes on my synths. Not for volume as much as I just like the way it sounds.

 

[legionserial]

Well, part of the stuff you hear lately about recording peaks at -14 or whatever has to do with giving yourself digital headroom so your tracks aren't clipping when you boost an EQ or something; but another big part of the reasoning is that 0dbFS is like 18db above the optimal level for a lot of analog gear. Pushing your preamps and stuff hot enough to reach 0dbFS can sometimes cause crappier sound as stuff overloads. Of course this part has nothing to do with your situation ITB; just thought you should know.

Oh I'm aware of that. My point was kinda the other way round. I meant that as the noise floor was so low you could afford to record it at a much lower volume. With the idea in mind that turning the track up in mixing isn't really going to make any noise on that track any louder because...well...there isn't really any noise anyway.

I usually mix my midi drums down drum piece by drum piece. With each piece peaking at about -12.5dbfs. What I'm trying to do is create the effect of an actual drumkit having been recorded. OF course is not exactly the same, I know that. So I want my audio files to be at the same level as they would be if I had actually recorded a drumkit. It might not be 100% neccessary to do it that way, but I kind of feel like its practice for if I was actually recording a real kit.

Of course I'm still kind of a beginner at this kind of stuff. So if what I have just said is completely wrong in anyway, feel free to give me a kick in the right direction

I don't have any problem turning my audio tracks down, knowing that I'm doing so in an effort to prevent overloading the mix buss.

What I meant about turning all the audio tracks down, was if you have 20 audio tracks, all at a pretty high volume, and 1 of them is not loud enough, your going to want to turn that track up. But if you haven't got enough room to turn that 1 track up enough without that individual track clipping because its already pretty loud, then your going to have to turn down all the other tracks instead. It just seems easier to have everything at a lower volume, so if that 1 track isn't loud enough, you can just turn that track up.

I say this becaue I used to track everything far too hot before I knew any better, and found myself having to turn down like 50 tracks just to get that 1 track to come through. Which I used to find annoying, and more time consuming than just turning that 1 track up. All my faders ended up being down really low, and once mixed, I had trouble pushing the master fader up high enough to get the whole tune at a decent volume. And I figure that if I have to have my master fader whacked up all the way, I may be doing something wrong.

I'm not sure about how it works with other software, but in Cubase, the faders go up to 6.02bdfs. So I figure that the absolute hottest I should track should be -(max you can put the fader up to)-0.1dbfs, ie in this case -6.12dbfs. That way if you do push the fader up to full with an audio track on it, that track ain't going to clip. It will be just below clipping. Of course I consider -6 to be a little loud for my tastes. I prefer to have the audio files so that if all the faders are set to 0 when I play it back, I won't have to turn the master fader too far down to avoid clipping. Or at least not as much as I used to have to. Which was a ridiulous amount.

Of course, all of this may have been mentioned in this thread before in some way that I may not have understood. As I say I'm still kind of a beginner with all of this, so this post is written as much in an attempt to be helpful, as it is to get verification that I am doing and understanding things correctly.

 

[Phuturistic]

I'm not sure about how it works with other software, but in Cubase, the faders go up to 6.02bdfs. So I figure that the absolute hottest I should track should be -(max you can put the fader up to)-0.1dbfs, ie in this case -6.12dbfs. That way if you do push the fader up to full with an audio track on it, that track ain't going to clip.

 

[legionserial]

If your sound file is peaking at -6.12dbfs with the fader set a at 0, then pushing the fader for that track up full (on Cubase at least) pushes it up by 6.02dbfs which = -0.1, which is just below clippng.

Screw it, I'm probably talking complete bollocks...

 

[Reggie]

Well, that is way hotter than you really need, but you probably won't be clipping the cubase mixer even if you boost the fader up. It runs on 32-bit floating, so it can sorta slide its scale up to accomodate levels over zero or however it does it. Now you may have plugins that will clip if you put them on that channel, but the Cubase mixer shouldn't.
But the master fader in Cubase is the one that shows you what the output of your final mixdown will be. Even it will sound OK with levels way over zero, but as soon as you mixdown to a stereo fixed-bit (24-bit or 16-bit) file you will hear the ugliness and damage you have caused.