How hot do you record in digital?

[The Audio Cave]

It's been reported by users that many digital consoles sound better when you bring the individual input levels down a bit. Mackie even recommended in the original d8b manual a tracking level of -15 or so. People complained it was harsh sounding but it sounds much better at those lower levels.

 

SX also sounds better to me when the tracking levels are down around-6 or so than up near 0 but that could be the clipping we don't see.

 

I'll say this again, even discussing mix buss levels in the context of how hot an individual track may or may not be is pointless. One has absolutely nothing to do with the other.

 

You could maximize every track of a 72 track project up to 0 and still mix them to peak your master bus at -15. Those tracks being hot have nothing to do with how you choose to mix the sum of those tracks into the master bus. Nothing.

 

 

This allows you to run the faders at a more comfortable level (-10-0db) without driving the mix summing into the red.

 

I don't care where my individual faders end up during a mix. A fader at -10 is no more comfortable for me than at -25 for a bass track or lead vocal or whatever. Why does it matter where the fader stops? Just push up a fader until the track is a the level where you want it. I don't get it.

 

 

You are a wise man to know to leave that masterfader at 0!!!!!!!

 

Pushing a master fader UP a little to make a song a little louder before printing a mix should not be a problem.

 

Lawrence

[Lee Knight]

 

Originally posted by The Audio Cave I don't care where my individual faders end up during a mix. A fader at -10 is no more comfortable for me than at -25 for a bass track or lead vocal or whatever. Why does it matter where the fader stops? Just push up a fader until the track is a the level where you want it. I don't get it. Lawrence

 

 

That's my thinking as well.

 

And... I've always made the assumption that having hot, individual track levels would never DEGRADE the sound. This is why I'm confused by some of these comments made earlier.

 

I guess if someone needs the fader in a certain range for better fader resolution (not audio resolution), as I understand some faders suffer from good resolution at lower levels.

 

I don't know. I don't get it either...

[wooden]

 

Originally posted by where02190 There's nothing wrong with that Lee, except that, especially at 24 bit, there's no need to track that hot. targeting nominal(average) levels to the 0dbu reference of your specific converters will provide you with an optimal audio quality signal, and allow plenty of headroom for transients without overshooting 0dbfs. This allows you to run the faders at a more comfortable level (-10-0db) without driving the mix summing into the red. You are a wise man to know to leave that masterfader at 0!!!!!!!

 

 

Can you explain the theory begind aiming the "digital 0dBu"?

[The Audio Cave]

 

Originally posted by Lee Knight And... I've always made the assumption that having hot, individual track levels would never DEGRADE the sound. This is why I'm confused by some of these comments made earlier.

 

 

It can degrade the sound if it's too close to zero because some of the peaks are actually over 0 but the meter doesn't show them. To get an accurate peak reading the software would have to recreate the file as if it's being converted back to analog. They don't. They read the level (peaks) of the sample points, not the top of the arc of the curves between the sample points which is the true peak. I'll refer back to the paper...

 

Digital Metering

 

Much more of a problem I think for mixing than tracking, the resulting mix file may have illegal overs that can't be read properly by many converters. Confusing stuff that Nika could explain better.

 

It's easy to leave a little headroom and not worry about it.

 

Lawrence

[Edge100]

 

Originally posted by Lee Knight That's my thinking as well. And... I've always made the assumption that having hot, individual track levels would never DEGRADE the sound. This is why I'm confused by some of these comments made earlier. I guess if someone needs the fader in a certain range for better fader resolution (not audio resolution), as I understand some faders suffer from good resolution at lower levels. I don't know. I don't get it either...

 

 

Faders are logarithmic, so you will have much better resolution (fader, not audio) at higher levels.

[Lee Knight]

 

Originally posted by The Audio Cave It can degrade the sound if it's too close to zero because some of the peaks are actually over 0 but the meter doesn't show them. It's easy to leave a little headroom and not worry about it. Lawrence

 

 

Right. I understand then. That is what I do. I've got plenty of headroom past my peaks. I just can't see recording in the bottom 3rd of my meters, you know?

[Kiwiburger]

Sorry - but I think there is a lot of unsubstantiated bogus rubbish being talked here.

 

Steinberg make it very clear that in Cubase SX or Nuendo, it is not possible to clip while in the 32 bit float domain. So this worrying about getting close, or even exceeding zero, or 'overdriving the mix bus' seems like a load of bollocks to me.

 

I have done some tests to check this out, and it's competely true. You make an audio file that greatly exceeds zero - and obviously this distorts like mad when played back in another audio editor. But re-import it into SX, and drop the master fader so the signal doesn't exceed zero - and the audio is completely undamaged.

 

I'm talking about Cubase SX, with the project set to 32 bit floating. So there is no conversions at any stage.

 

I think the acid test would be to record some pure sinewaves, and prove what happens when the 'mix bus is overload' - but I am confident that there is no problem.

 

My experience with Logic 5 (before the bastards killed Logic) is that - all other hardware factors being equal - Logic audio quality was inferior to Cubase SX, so i don't even bother testing Logic any more. My ears tell me something isn't right. The mix bus and - more importantly the effects bus - of protools is always being debated, so if you tell me there are problems exceeding zero in Protools I will easily believe that.

 

But Nuendo/Cubase sound better - to me and many other people - and maybe this subject has something to do with it?

[The Audio Cave]

Okay this is gonna take more than a minute...

 

Steinberg pushes recording and rendering true 32-bit float files for a good reason but they advertise it the wrong reason. It's true that it will not clip in SX when working on a 32-bit project. But that's not the main benefit of working with a 32-bit project. Although I don't use 32-bit projects in SX it may be a good idea.

 

There is an inherent problem with all native daws that do 32-bit float internally. When the audio leaves the mix bus it has to be 24-bits because there are no 32-bit converters. They discard 8 bits. Those aren't just 8 zeros, there is "relevant" data in some of those bits.

 

ProTools is the same. The 48-bit HD or 32-bit float LE has to be 24-bits when it leaves the software to interface with other digital gear or analog converters. The best way to do that is to apply some kind of 24-bit dither to mask the truncation errors and smooth things over.

 

HD does that automatically I think. Native apps that use 32-bit float do not. Why? Because, to quote people smarter than me, "...dither is not as effective with floating point math...", so they let the user decide if they want to dither the mix bus or not. If you don't? You get tiny truncation errors that can build up and eventualy may start to be heard.

 

By rendering a SX "in the box" mix to a true 32-bit float file (or "writing" 32-bit float files if you render a track) you avoid those errors in the track and master mix file. What does that mean?

 

It means you can import that file back into SX later and edit, master, add dsp, whatever and you keep it at the full fidelity of the SX 32-bit internal audio engine of SX with no truncation errors. When you get ready to burn your CD file, dither it down to 16-bits and it should sound pretty good. It also means much greater headroom.

 

What happens when you use a 24-bit file the same way, bringing it in and out of SX or another app going from 24 to 32 over and over again adding tiny errors at every step?

 

Every time you bring a file into an app like SX or Wavelab or Sonar etc..., it becomes 32-bit float again in the software. You add plugs, eq and you make a new 24-bit file with the tiny truncation errors. Eventually those errors will build up and may affect the audio. A little harsher? A little less open?

 

Rendering a true 32-bit float file as a master mix preserves the entire bit depth of the internal audio engine and should Theoretically be "better" because it never get's truncated or dithered. What you hear through the speakers does because you have to convert it to 24-bits to hear it, but the files never get truncated. Those tiny errors never get into the audio files.

 

So (theoretically) when you finally dither that thing to 16bit for CD it is "the best it can be".

 

That's the main benefit of "recording" at 32-bits in SX although it only makes sense if you're mixing in the box.

 

The question is... do you want to give up the additional disk space and is it worth it? Let your ears be the judge.

 

Lawrence

[where02190]

I'm not a mathematition so no I cannot explain the theory, all I know is that not only myself, but all the engineers I know hear the difference in the quality of the mix. Protools, DP, Nueno, etc, converters from Apogee, M-Audio, Mackie, Edirol...the results to our ears were always the same, keep the input to the mix buss at the converters nominal 0dbu levels, and the mix sounds vastly better than pounding into the mix buss and lowering the master fader. More open, better clarity, and much punchier.

 

I don't care about numbers, I care about the end result sounding great.

[gsHarmony]

 

It can degrade the sound if it's too close to zero because some of the peaks are actually over 0 but the meter doesn't show them. To get an accurate peak reading the software would have to recreate the file as if it's being converted back to analog. They don't. They read the level (peaks) of the sample points, not the top of the arc of the curves between the sample points which is the true peak.

 

 

This is an extremely good point. Actually, come to think of it, this means that normilization could be very evil too. I guess CD players could start using D/A converters capable of outputing a greater voltage range, which would probably solve the problem, but I don't think this will happen. Are there even any tools available that actually show the peak of the reconstructed analog signal?

[The Audio Cave]

 

...keep the input to the mix buss at the converters nominal 0dbu levels, and the mix sounds vastly better... than pounding the mix bus...

 

 

I agree totally because you're absolutely right. We're saying the same thing. I'm not a mathemetician either.

 

I agree 100%.

 

Lawrence

[The Audio Cave]

 

Originally posted by gsHarmony This is an extremely good point. Actually, come to think of it, this means that normilization could be very evil too. I guess CD players could start using D/A converters capable of outputing a greater voltage range, which would probably solve the problem, but I don't think this will happen. Are there even any tools available that actually show the peak of the reconstructed analog signal?

 

 

Give the credit to Nika Aldrich. My brain could have never grasped that concept.

 

Anyway...

 

I think that DSD systems do that. That's one reason why people say it sounds so good, or so I'm told. My understanding is that it will not allow the mastering engineer to maximize it to the point of illegal overs. It just doesn't allow it.

 

You'd have to ask Brad Blackwood maybe about the specifics of that.

 

Lawrence

[Zooey]

 

Originally posted by The Audio Cave Give the credit to Nika Aldrich. My brain could have never grasped that concept.

 

 

He's written more on the exact topic of tracking levels on the Digidesign DUC and other forums. You asked earlier if tracking less hot compromised your audio. The answer is no.

 

A track that peaks at -18dBfs captures the same dynamic range of any real world source as a track that peaks at digital zero. If you were to normalize the two tracks to the same level and compare them back to back, both tracks would represent the dynamic range of the recorded source with identical accuracy. It's complicated stuff, but you can look it up.

[wooden]

Originally posted by The Audio Cave I agree totally because you're absolutely right. We're saying the same thing. I'm not a mathemetician either. I agree 100%. Lawrence

 

Thanks guys, I am not a mathematician either, but i do know some thing.

 

I have been shooting as close of the 0 dBFS as I can, and my mixes have sounded saturated, almost like over compressed, and "empty", and it could be the cause... very eye opening.

 

But, in order to get that you dont necesarily have to keep your preamp at 0dBU ? right? do you?

[Kiwiburger]

We seem to be arguing and agreeing at cross purposes - i'm confused.

 

TheAudioCave - you are expressing exactly what I believe and do. I record with 24 bit converters into SX. From that point on, the audio is in 32 bit floating. All internal audio, plugins etc are in 32 bit floating. I keep the project at 32 bits - all audio exports and re-imports are at 32 bit floating. Although i'm monitoring through 24 bit converters, my final mix, and self masters never go through converters ever. I mix down to CD audio, and that is the only time I apply dithering.

 

So where is this mix bus that supposedly can overload? The SX mix bus can't clip. I set my track faders as high as I want, and reduce my master fader for any audio exports. I can't hear any problem. Because i'm avoiding conversion, my audio is cleaner than a lot of commercial stuff.

 

I can totally agree that everything changes as soon as analog comes into the picture - and A/D and D/A converters are included in that.

 

Totally agree that dynamic range is not lost at lower settings - because the dynamic range of 24 bits is excessive anyway.

 

Totally agree that you can 'track too hot' - and this will be because of the weakness of the A/D converter in that region.

 

But between the A/D and the D/A - assuming no conversions and no dither, I can't see how digital that doesn't clip can ever be 'too hot'.

[where02190]

Originally posted by wooden Thanks guys, I am not a mathematician either, but i do know some thing. I have been shooting as close of the 0 dBFS as I can, and my mixes have sounded saturated, almost like over compressed, and "empty", and it could be the cause... very eye opening. But, in order to get that you dont necesarily have to keep your preamp at 0dBU ? right? do you?

 

Nope. Just adjust the output level of the pre to adjust the level hitting your converter. I drive the hell out of pres sometimes intentionally to get a specific sound, and either turn down the output, or of they don't have one, I'll put a clean comp inline and just use it as a gain stage, and bypass the compressor(turn the threshold all the way up and put the ratio at 1:1). This gives me the comps output gain to compensate for the hot preamp output. (OUr TL Audio PA-2's get this treatment alot, having no output level control.)

[wooden]

Thanks, just one more thing. Where are talking about aiming to the "digital 0 dBU" rms or peak?

[The Audio Cave]

 

Originally posted by Kiwiburger We seem to be arguing and agreeing at cross purposes - i'm confused. TheAudioCave - you are expressing exactly what I believe and do. I record with 24 bit converters into SX. From that point on, the audio is in 32 bit floating. All internal audio, plugins etc are in 32 bit floating. I keep the project at 32 bits - all audio exports and re-imports are at 32 bit floating. Although i'm monitoring through 24 bit converters, my final mix, and self masters never go through converters ever. I mix down to CD audio, and that is the only time I apply dithering.

 

 

I hear you. That seems like the best way to go. If what you say is true (and I believe you) then pulling the master fader down must be lowering the pre-fader level going into the mix bus. I was curious about that.

 

Cool about the 32-bit rendering. I'm going to try that on my ITB mixes.

 

Thanks.

 

Lawrence

[Zooey]

Originally posted by wooden I have been shooting as close of the 0 dBFS as I can, and my mixes have sounded saturated, almost like over compressed, and "empty", and it could be the cause... very eye opening.

 

That could be the sound of the analog portion of your gear (preamps, etc.) being driven far outside of its optimal range. It may not register to your ears as "distortion," but the cumulative effect is what you describe.

 

The first time I tracked closer to 0VU, I was convinced that I had ruined the recording. The waveforms looked so...wimpy. That doesn't mean {censored}. Many DAW users should break themselves of the habit of obsessing about how their audio looks and start focusing on how it sounds.

[wooden]

 

Originally posted by Zooey That could be the sound of the analog portion of your gear (preamps, etc.) being driven far outside of its optimal range. It may not register to your ears as "distortion," but the cumulative effect is what you describe. The first time I tracked closer to 0VU, I was convinced that I had ruined the recording. The waveforms looked so...wimpy. That doesn't mean {censored}. Many DAW users should break themselves of the habit of obsessing about how their audio looks and start focusing on how it sounds.

 

 

And what would be your advise for me?

[greendoor]

Get to know your gear. There is a big difference in analog thinking and digital thinking.

 

A lot of analog preamp and compressors have VU meters. VU meters use about half the meter for the last 6dB. So if you want to learn how it sounds at much lower levels, you might not even see much meter movement. Same with digital peak meters and waveforms - you might need to get used to not seeing much movement or wimpy looking waveforms.

 

Some preamps sound better when cranked - others don't. I don't think there is any magic formula - just learn what it sounds like and use what works.

[wooden]

Hey audio cave or where. Jus one more question. Do we aim to the digital "0 dBU" meterin with peak or rms?

[where02190]

 

Originally posted by wooden Hey audio cave or where. Jus one more question. Do we aim to the digital "0 dBU" meterin with peak or rms?

 

 

See your specific converters specs.

[guitarbth]

Just to make things frustrating... what if your preamp doesn't have a VU meter? Are there any software VU's that I can use to reference the dBU level along with the dBFS meter that my DAW has?

 

Brandon

[guitarbth]

Nevermind... found one - free software VU meter.... PC and Mac

 

http://www.pspaudioware.com/

 

Brandon