Best Recording Levels

[PTUser NYC]

The manual is wrong, and I have discussed this with many big name Grammy award winning engineers, and I find that they mostly all agree with my assertions.

Headroom is a good thing, and driving your analog side harder than you would if you were using analog makes no sense.

The noise floor in digital is SO far below the noise in your signal that your premise is ridiculous.

Its OK. We don't need to decide between us. People can read, experiment and do what they like.

All sorts of things conspire to create the sound of our work, and people will judge us on results. If you get good results working hot, more power to you.

[PTUser NYC]

Originally Posted by mightyduck

That's not correct. Johnson noise and quantization distortion have different characteristics. Quantization distortion tends to be much more intrusive, and is not necessarily "masked" by analog noise.

Please explain the application of dither then. Isn't is noise precisely meant to mask quantization error? Further, we can say that the Quanitzation error is half the size of the quanta, so we know that in a full scale recording, it's at -144db, and if you record 18 db lower as I'm suggesting, then its at -126db, and still WELL below the noise in the converter, or your signal, which also acts like mega dither in this case.

Originally Posted by mightyduck

Also, you seem to be completely ignoring sounds that are synthesized within the daw, thus having no analog noise component.

And you don't set a recording level for them either. Software synthesizers have plenty of noise anyway.

Originally Posted by mightyduck

It seems to me that your figures are somewhat "doctored" there. Using pro equipment you are just not going to be overdriving stuff in the way you describe.

You will be eating up headroom.

Originally Posted by mightyduck

You seem to be saying that the meters on a daw are to be read the same way as a VU meter. That's not correct. DAW meters are generally peak meters. Peaks of -6dBfs or so may generally give you the -18dBfs RMS reading that you want to correspond to 0VU.

I'm advocating observing 0dbVU. If peaks occur at -6dbFS, so be it. Usually, they are lower.

Originally Posted by mightyduck

Also, as you noted in a subsequent post, there are trim controls, and the solution is not to record unduly low, but, rather to adjust the trims on your interfaces to optimize both digital level and analog level. My 192 interfaces came from the factory set at -15dBfs = 0 VU. Rental houses will typically calibrate to -14dBfs or -15dBfs for music work. Film work can be a special case on a soundstage, or live field recordings, where you may be faced with very unpredictable loud sounds, and -20dBfs is common there.

I do agree that you can address some of what I'm talking about by calibrating your interface. I still think headroom is a good thing in the PT mixer.

Originally Posted by mightyduck

Nothing's really changed. Its just a couple of bits better. By the way, you don't get 24 bits, maybe 19-21, or so.

Well, so to speak. I did say all the way through my argument that the noise floor in the converter was higher than the 24 bit limit. I used 120db, which is analogous to "19 bit" in your slang. It's still 24 bit, just with a higher noise floor.

I was being conservative with peaks at -18dbFS to show that even that low, the noise isnt a problem. I'm instead advocating levels around -18dbFS, with peaks above that. But I'm certainly not saying to record up near the top of the meter.

Originally Posted by mightyduck

That is a complete non-issue. Most plug-ins have input trims, or you can simply back the signal down in the 48 bit mixer after recording it at an optimal level [or as optimal as is reasonably practical in the circumstances].

Sure, but now your changing your interfaces operating level, and having to do all this gain chasing all throughout your chain. Why not just work easy with lots of headroom?

I still think the summing sounds better with some headroom.

Originally Posted by mightyduck

Most of what you are cautioning against is easily overcome by simply knowing how to record and mix. The "bad results" from "recording too hot" that we are always hearing about is simply people doing stupid stuff.

Yes, there are many ways to skin a cat, but for this forum, for people who are asking for advice on levels, I think recording lower is a better answer.

I can think of one mix engineer who used to mix major radio hits in the 80s. He was overdriving EVERYTHING in the studio, and his records sounded great. Different engineers have different techniques, and ways to make the sounds they get work in context of other sounds they get.

That said, there's No reason to record hotter, and at BEST you have to jump through hoops to ameliorate the damage. And that's only IF you don't think headroom in summing matters.

And all that for no audible benefit.

Originally Posted by mightyduck

As explained above, you seem to be confusing peak meters with VU meters.
There is no good reason to record with peaks at -18dBfs. That's unduly low in most cases.

I understand the difference. I used peaks to explain the noise floor because its easier to discuss mathematically.

Lastly, this is a very interesting discussion/debate. I mean no disrespect, and thanks for the discourse.

[mightyduck]

Originally Posted by PTUser NYC

Lastly, this is a very interesting discussion/debate. I mean no disrespect, and thanks for the discourse.

No problem. But I can assure you that my premise is not "ridiculous".

As I stated, I think you have some good information in your post, and I'm not in this for a "fight".

I'll respond further to some of the issues a little later, as I have a lot to do right now. I agree that we can just present the information and others can check it out, make use of it, or not, as they feel it applies to them.

I should clarify one thing quickly. You mentioned "recording hot". I am not talking about "recording hot". I am talking about doing your initial sampling at or near the optimum for the adc. Recording "hot", or "mixing hot" often implies overdriving something, or clipping something, and that is not what I am posting about. The daw is just like any other component in the recording chain. It has input and output trims that allow you to optimize its internal levels, while allowing for the external analog [or digital] equipment to also operate at its optimum levels.

[O.G. Killa]

Originally Posted by mightyduck

I am not talking about "recording hot". I am talking about doing your initial sampling at or near the optimum for the adc. Recording "hot", or "mixing hot" often implies overdriving something, or clipping something, and that is not what I am posting about. The daw is just like any other component in the recording chain. It has input and output trims that allow you to optimize its internal levels, while allowing for the external analog [or digital] equipment to also operate at its optimum levels.

some people just don't get it.

[PTUser NYC]

The main point of contention here is whether Quantization error is masked by noise. I say yes, and mightyduck says no. I say yes becasue that's what dither does.

Further we can quantify the level of the error. It's half a quanta.

Even with peaks at -18dbFS (and I'm advocating a level there, not peaks) the Q error is down below -126 db - well within the noise floor of the converter, not to mention the signal itself which never EVER has 126 db S/N.

It's simply a non issue.

Yes, some of my arguments against recording higher than this can be addressed by recalibrating your converter, and using trims throughout mixing. But WHY? Why not just operate arou8nd 0dbVU as was intended.

[mightyduck]

Originally Posted by PTUser NYC

The manual is wrong, and I have discussed this with many big name Grammy award winning engineers, and I find that they mostly all agree with my assertions.

When I see "the manual is wrong", it naturally gives me pause. What is in the Pro Tools Manual is confirmed in just about every other text or manual around, and is also the opinion of guys like Paul Frindle and other designers [people who are, by the way, well aware of your "analog noise floor argument", which is rather old news in these type of discussions].

Moreover, the "I talk to Grammy-winners" angle is also a give-away. I know you stated you didn't want an argument, but with your recent post quoted below, and the "Grammy-winning engineers" stance, it seems you may be a bit more emotionally invested than you are letting on.

Are these "Grammy-winners" for "Best Engineered Album on Pro Tools" or some other engineering feat. Or are they just guys that happened to engineer records that won in other categories. A lot of pretty funny Grammies have been awarded, you know.

Originally Posted by PTUser NYC

The main point of contention here is whether Quantization error is masked by noise. I say yes, and mightyduck says no. I say yes becasue that's what dither does.

Further we can quantify the level of the error. It's half a quanta.

Even with peaks at -18dbFS (and I'm advocating a level there, not peaks) the Q error is down below -126 db - well within the noise floor of the converter, not to mention the signal itself which never EVER has 126 db S/N.

It's simply a non issue.

Yes, some of my arguments against recording higher than this can be addressed by recalibrating your converter, and using trims throughout mixing. But WHY? Why not just operate arou8nd 0dbVU as was intended.

I stated that I would address some of the issues raised when I have time to see to it.

It would be preferred if you don't try to state my position for me, especially in an spin attempt to frame the issues. I can speak for myself,

There are a number of issues aside from the one you mentioned.

Anyhow, just to clarify a couple of points briefly, the magnitude of the quantization error is not "half a quanta", its half or less. Dither does not "mask" quantization distortion, its function is to prevent it. With all due respect, I think your concept of dither may need some work.

I note that you have kind of walked some of your previous statements back to say that you are not necessarily "advocating" peaks at -18dBfs. So perhaps we have more agreement than was previously evident. But level choices are often best made on a "case-by-case" basis, and any attempt to impose a rigid standard in the current state of affairs is bound to be problematic.

Anyhow, I'll post some further information later, that some people may find helpful or interesting, if they have not seen it before in some of the other rather long-winded threads on this topic.

[PTUser NYC]

OK I'm not going to fight with you.

I'll just say this

1) The Grammy award winners I'm talking about are amazing world class engineers. I'm speaking of several, and I won't name them here, but you would also agree that they are fantastic and know what they're talking about. They agree with what I've said too.

2) I'm not walking back any statements. I said peaks at -18dbFS for the purpose of calculating dynamic range. Had I used average level, we'd be arguing about where the peaks are.

There is a difference between the math I'm using to calculate dynamic range, and the levels I'm suggesting.

3) The quantization error IS measurable at half a quanta. If it were MORE, then it would have rounded to a different quanta.

4) The manual IS wrong. It just is.

There's NO reason to record hotter than 0dbVU into Pro Tools.

[O.G. Killa]

here are a some posts from Paul Frindle on the subject...

http://www.gearslutz.com/board/5063002-post71.html
http://www.gearslutz.com/board/5063003-post72.html
http://www.gearslutz.com/board/5063007-post76.html
http://www.gearslutz.com/board/5063050-post118.html
http://www.gearslutz.com/board/5734741-post753.html
http://www.gearslutz.com/board/5656497-post3656.html
http://recforums.prosoundweb.com/ind...5c28ada4873c68 (scroll down near the bottom)

And another thread...

http://recforums.prosoundweb.com/index.php/t/15038/0/

[O.G. Killa]

Originally Posted by PTUser NYC

There's NO reason to record hotter than 0dbVU into Pro Tools.

The important thing to realize here is that VU is an averaging meter. 0dBVU on a Tamborine can easily hit -6dBFS inside your daw, while a bass guitar hitting 0dBVU would probably never go above -18 or -17dBFS in the DAW. Recording something like a Snare drum with it's peak at -18dBFS might show up as -5 or -8 on a VU meter, not 0.

So long as you calibrate using a 1KHz sine wave at +4dBu=0VU= -18dBFS (or -20dBFS) = 85dB SPL, you never really need to worry about the levels because your analog AND digital gear will start clipping at the same time (around 9 Volts RMS/30 Volts Peak to Peak, give or take) and it will start to sound painfully loud at the listening position before it even gets close to clipping.

The problems arise when people turn down their monitoring level (or not calibrate their monitors as part of the signal chain at all) and then TRY to record as hot as possible with every instrument.

So, a tamborine hitting -2dBFS might be +3VU, but a Bass guitar hitting -2dBFS is probably around +16VU!! The tamborine will sound fine, the bass guitar will have harmonic distortion and "digititus".

Then there is the issue of noise. Noise does mask other noise. In fact, any sound can mask any other sound if the difference in level is great enough. The closer in frequency the two sounds are, the less difference in volume is needed for masking. This means that "analog" noise from the converter at -120dBFS will mask quantization noise of the system down at -144dBFS. But this is irrelevant as you will see in a second.

If your best piece of gear only has a dynamic range of 100dB, that means from clipping (+22 or +24dBu) down to the noise floor is 100dB. This means with the piece of gear turned all the way down, it will still have 20dB MORE noise in it than your converter (which has noise 120dB down from clipping). Turning the gain on this piece of gear UP, raises the noise floor and decreases the dynamic range. So, for something like a vocal (especially if the vocalist is a soft singer), trying to record as hot as possible without clipping the converter is going to raise your noise floor up to somewhere around -70dBFS to -40dBFS. And that is going to make one heck of a noisy/hissy recording (and the linearity of most devices goes out the window at the top of it's dynamic range). Not to mention what's going to happen when you shrink the dynamic range even more with compression on the vocal and compression on the mix. Now your -70~-40dB noise floor gets raised up to -50~-20dBFS.

Now what happens when you convert to 16bit? The noise floor of the digital system raises from -144dB up to -96dBFS. Anything below -96dBFS is lost and dither removes any quantization distortion by raising the noise floor slightly. so you noise floor is now -93dBFS or so. BUT, the noise of your vocal mic is up around -50dB~-20dBFS and so is still very much present in the 16bit master since it is 37dB ~ 67dB LOUDER than the new noise floor.

But if you were to keep your levels lower and not record as hot as possible, you would still have all the clarity of your orginal signal, but the noise of the preamp would most likely be removed when switching to 16bit and Dithering, leaving you with a much cleaner, more pristine recording that has a noise floor of -93dBFS, not -50dBFS or up to -20dBFS.

And this isn't even dealing with what happens to hot signals within the plugins or at the output of the DAC. Which is another issue all together.

My best piece of advice to anyone and everyone. Start your calibration process with your MONITORS and work backwards. Once your listening environment is calibrated so that +4dBu (pink noise) = 85dB SPL, your ears will tell you what is too loud and what is too soft. And you'll notice that calibrating the rest of your equipment using a 1KHz sine at +4dBu makes sense and yields great sounding recordings.

[Dism]

Originally Posted by O.G. Killa

The important thing to realize here is that VU is an averaging meter. 0dBVU on a Tamborine can easily hit -6dBFS inside your daw, while a bass guitar hitting 0dBVU would probably never go above -18 or -17dBFS in the DAW. Recording something like a Snare drum with it's peak at -18dBFS might show up as -5 or -8 on a VU meter, not 0.

So long as you calibrate using a 1KHz sine wave at +4dBu=0VU= -18dBFS (or -20dBFS) = 85dB SPL, you never really need to worry about the levels because your analog AND digital gear will start clipping at the same time (around 9 Volts RMS/30 Volts Peak to Peak, give or take) and it will start to sound painfully loud at the listening position before it even gets close to clipping.

The problems arise when people turn down their monitoring level (or not calibrate their monitors as part of the signal chain at all) and then TRY to record as hot as possible with every instrument.

So, a tamborine hitting -2dBFS might be +3VU, but a Bass guitar hitting -2dBFS is probably around +16VU!! The tamborine will sound fine, the bass guitar will have harmonic distortion and "digititus".

Then there is the issue of noise. Noise does mask other noise. In fact, any sound can mask any other sound if the difference in level is great enough. The closer in frequency the two sounds are, the less difference in volume is needed for masking. This means that "analog" noise from the converter at -120dBFS will mask quantization noise of the system down at -144dBFS. But this is irrelevant as you will see in a second.

If your best piece of gear only has a dynamic range of 100dB, that means from clipping (+22 or +24dBu) down to the noise floor is 100dB. This means with the piece of gear turned all the way down, it will still have 20dB MORE noise in it than your converter (which has noise 120dB down from clipping). Turning the gain on this piece of gear UP, raises the noise floor and decreases the dynamic range. So, for something like a vocal (especially if the vocalist is a soft singer), trying to record as hot as possible without clipping the converter is going to raise your noise floor up to somewhere around -70dBFS to -40dBFS. And that is going to make one heck of a noisy/hissy recording (and the linearity of most devices goes out the window at the top of it's dynamic range). Not to mention what's going to happen when you shrink the dynamic range even more with compression on the vocal and compression on the mix. Now your -70~-40dB noise floor gets raised up to -50~-20dBFS.

Now what happens when you convert to 16bit? The noise floor of the digital system raises from -144dB up to -96dBFS. Anything below -96dBFS is lost and dither removes any quantization distortion by raising the noise floor slightly. so you noise floor is now -93dBFS or so. BUT, the noise of your vocal mic is up around -50dB~-20dBFS and so is still very much present in the 16bit master since it is 37dB ~ 67dB LOUDER than the new noise floor.

But if you were to keep your levels lower and not record as hot as possible, you would still have all the clarity of your orginal signal, but the noise of the preamp would most likely be removed when switching to 16bit and Dithering, leaving you with a much cleaner, more pristine recording that has a noise floor of -93dBFS, not -50dBFS or up to -20dBFS.

And this isn't even dealing with what happens to hot signals within the plugins or at the output of the DAC. Which is another issue all together.

My best piece of advice to anyone and everyone. Start your calibration process with your MONITORS and work backwards. Once your listening environment is calibrated so that +4dBu (pink noise) = 85dB SPL, your ears will tell you what is too loud and what is too soft. And you'll notice that calibrating the rest of your equipment using a 1KHz sine at +4dBu makes sense and yields great sounding recordings.

+1

Speakin truth as always!