Keeping your levels low in digital recording/mixing...

[where02190]

Of course, you don't have to believe anything you read on the net. Try it both ways and see which way sounds best to your ears.

And this is what it all comes down to. (BTW music-tech, nice post, well written.)

The bottom line is all the math in the world doesn't make a good sounding recording if it doesn't sound good. Over and over engineers both renounded and newbies hear the difference when proper gain staging and 0dbu converter levels are attended to. Digital is very unforgiving, and doen't, like it's analog predecessor, have a "sound" that's anything remotely musical when the red lights are lit.

I think this post is one of the best I've ever seen on this or any forum.

[Peel]

Ah, music-tech, perfect post that was.

[music-tech]

Thanks guys! Glad you found the post helpful.

I think it was a fairly common misconception, in the earlier days of digital, that you needed to keep all your digital levels as close to 0dBFS as possible to maximize the "resolution" and get the best sound out of digital. I was guilty of believing that as well, and it's a really hard habit to break after you've been pushing things up to the brink of clipping for a long time. I still find myself wanting to push those levels up subconsciously, and having to remind myself that it's not going to sound better that way, and may potentially sound worse (even though I have great converters with lots of headroom on the analog side, it's still best not to push them too hard).

Maybe if converter and software manufacturers started making those digital meters with a display that shows the analog reference 0dBVU levels, and any lights above that are yellow and red, so that you know you are pushing the levels above reference levels, then it would be easier for people to record at more normal levels. The way it is now on most converters and software meters, the signal level lights only turn red once it's too late and you're already clipping.

[teacue]

Possibly a good time to mention the wonderful VST Peak Programme Meter "PPMulator+" from Raw Material Software.
It does indeed help a lot to get the right level
http://www.rawmaterialsoftware.com/ppm.php

[jupiter8]

I do agree that music-techs post was valid,to the point and in general really good. However it does nothing to support the notion that MIXING at low volumes somehow magically transforms your mixes into sounding better. Au contraire he mentioned nothing to even remotely support that theory.

At what level you should hit you AD and DA converter is a completly different subject and is a whole other topic initself.

The truth is you can still mix with every track hitting close to zero and just lower the master fader and you will get the EXACT same result as if you lowered the individual channels.

Of course you should trust your ears to a point but be carefule so you dont enter John Vestman territory.
Math's and science may not be the way to great sounding mixes but they're very effective at shattering ridicoulus myths that "pros" who really don't know what they're talking about are spreading. Or maybe they're discussing a completly different platform (integer) which very few of us use. It is good to take advice from succesful names but just because they've sold shitloads of records does'nt mean it applies to your situation or even that they're right.

[music-tech]

I do agree that music-techs post was valid,to the point and in general really good. However it does nothing to support the notion that MIXING at low volumes somehow magically transforms your mixes into sounding better. Au contraire he mentioned nothing to even remotely support that theory.

EXACTLY! I have no opinion on the theory that mixing at lower levels will somehow magically make your mixes sound better, as I've never tested it for myself. As I said in my post, this comes down to the programmers and how well they implement the mix engine and the math. If they do the math correctly, one would think that it should make no difference to the sound. However, as I also mentioned in my post, not all plug-ins are floating point, and if you are driving your channels too hard and a fixed point plug-in is being clipped, then that can definitely affect the sound.

The only other possible reason I can even think of why mixing with all channels at lower volumes would have any effect on the sound starts getting into the pros and cons of floating point versus fixed point. I'm probably not very good at explaining this, and not even sure if I understand all the details completely of how it works. However, the con of floating point math is that your digital noise floor (from quantization/rounding errors), is NOT at a fixed level when using floating point math, and the level of that noise modulates as the floating point multiplier changes values. Some people claim that it makes the digital noise floor more audible when it modulates, as opposed to having that constant level you get with fixed point math... although, it's still at a VERY low level either way and I personally don't think most people in most normal mixing would ever be able to hear it, or its effects, at all. But, maybe when you add up a whole bunch of channels being driven hard, the cumulative effect of all that modulating noise is somehow becoming more audible. This is really just a wild guess at what MAY possibly be some sort of explanation, although I still think it's a bit far fetched.

However, I will point out that I mix through a Yamaha O2R96 digital mixer, which uses 48 bit fixed point processing, with 56 bit accumulators. When I'm composing "in the box", I usually just have all the tracks coming out of one stereo bus into the board for monitoring. Once I start getting ready to mix, I split out all the tracks to individual outputs going to the Yamaha board. Without changing any levels, and with all the faders on the O2R96 at unity gain, I have often thought that as soon as I split things out, everything started sounding better. But, this is totally unscientific, and I've not tried to verify it in any way. Perhaps I like the "sound" of the fixed point mixing engine in the Yamaha O2R96 better than the floating point mix engine in Nuendo?? Or perhaps the panning laws are slightly different between Nuendo and the O2R96, and so there is some slight volume changes on each channel that fools me into thinking it sounds better.

Basically, I mix through the O2R96 for convenience of working with real faders and being able to interface with my external effects units easier, and it sounds good enough for me. I have also done some quick projects for low budget jingle clients completely in the box to make it easier to quickly render all the different mix variations they need, and that all sounds fine also... and I have driven those levels hard on each channel, as well as driving the master bus hard with a limited at the end of the chain to get a finished "mastered" mix (a big "no-no" I know, but when you have super low budget clients, you have to work fast)... and, it still sounded great and they were happy.

[Peel]

Yes, well I liked music-tech's post exactly because it stuck clearly to the case -- tracking -- where keeping your levels low (well, a lot lower than most people think is ideal) is good for sound quality. So much of this never-ending back & forth comes from neglecting to differentiate between very different cases (both from simple ignorance and from sloppy writing / reading).
Yeah, the slashed "recording/mixing" in the topic creates some unhelpful ambiguity, but "at what level you should hit your AD converter" falls under the heading of "recording," I think.
I, (and I'm sure music-tech as well) agree with the rest of what you and all sensible people say about the all-inside-the-box case.

[Mr. Tingle]

Am I to assume that one is still to try and get the signal-to-noise ratio as high as possible on the microphone preamp? After all there is the noise from the microphone and preamp, you want that as quiet as possible.

If the loudest point of the recording hits just below 0db on the convertors and then as long as the levels on the master fader don't go above 0db (and any vsts are programmed correctly) then this will be the 'best' quality available?

[bduffy]

Am I to assume that one is still to try and get the signal-to-noise ratio as high as possible on the microphone preamp? After all there is the noise from the microphone and preamp, you want that as quiet as possible.

If the loudest point of the recording hits just below 0db on the convertors and then as long as the levels on the master fader don't go above 0db (and any vsts are programmed correctly) then this will be the 'best' quality available?

Personally, I like to keep my peaks between -6 to -3dB, to have some leeway.

[where02190]

I do agree that music-techs post was valid,to the point and in general really good. However it does nothing to support the notion that MIXING at low volumes somehow magically transforms your mixes into sounding better. Au contraire he mentioned nothing to even remotely support that theory.

This and many, many other posts and articles do support this to be true, that the quality of audio is very noticably better when proper gain staging is observed.

At what level you should hit you AD and DA converter is a completly different subject and is a whole other topic initself.

Completely false, and it is specificially related. Again, the result of proper gain staging throughout.

The truth is you can still mix with every track hitting close to zero and just lower the master fader and you will get the EXACT same result as if you lowered the individual channels.

Again completely false, and a trained ear knows this.

Of course you should trust your ears to a point but be carefule so you dont enter John Vestman territory.

All the math in the world means nothing if it doesn't sound good. We're not mixing to satisfy a formua, we're mixing for the human ears.

Math's and science may not be the way to great sounding mixes but they're very effective at shattering ridicoulus myths that "pros" who really don't know what they're talking about are spreading. Or maybe they're discussing a completly different platform (integer) which very few of us use. It is good to take advice from succesful names but just because they've sold shitloads of records does'nt mean it applies to your situation or even that they're right.

This is just funny. The pros "don't know what they're talking about". Why do you think they are pros? Because they DO know what they are talking about. Many of them are hands on in development of the technology that is part of the digital process today.

Bottom line, it's what the music sounds like, not what it looks like or what the math says it will be. You obviously can't hear those differences. Engineering is a tallent, and we do not all possess it.

[jupiter8]

I do agree that music-techs post was valid,to the point and in general really good. However it does nothing to support the notion that MIXING at low volumes somehow magically transforms your mixes into sounding better. Au contraire he mentioned nothing to even remotely support that theory.

This and many, many other posts and articles do support this to be true, that the quality of audio is very noticably better when proper gain staging is observed.

No it does'nt. He even said so himself unless you did'nt notice.

At what level you should hit you AD and DA converter is a completly different subject and is a whole other topic initself.

Completely false, and it is specificially related. Again, the result of proper gain staging throughout.

You're just saying i'm wrong yet again fail to even remotely back up your standpoint. Saying i'm wrong does'nt make it so.
Hitting the AD converters will of course affect the sound as they are analog and you will get different resolutions. How is that even remotely related to what level your faders are ?
Or do you mean that changing the levels is bad ?

The truth is you can still mix with every track hitting close to zero and just lower the master fader and you will get the EXACT same result as if you lowered the individual channels.

Again completely false, and a trained ear knows this.

This is so easy to prove and i even showed you how to do it.Why are you so afraid to try it out for yourself ? It's dead easy. You do realize that expecting something to sound different will make you believe it will sound different.However a quick blindtest will reveal it ain't so.
Or do you believe that 2 soundfiles that cancel eachother out when one is phase reversed can sound different ?

Of course you should trust your ears to a point but be carefule so you dont enter John Vestman territory.

All the math in the world means nothing if it doesn't sound good. We're not mixing to satisfy a formua, we're mixing for the human ears.

So John Westman claims he can hear the difference between different brands of harddrives. Just because he has sold more records than me does that mean he is right ?

Math's and science may not be the way to great sounding mixes but they're very effective at shattering ridicoulus myths that "pros" who really don't know what they're talking about are spreading. Or maybe they're discussing a completly different platform (integer) which very few of us use. It is good to take advice from succesful names but just because they've sold shitloads of records does'nt mean it applies to your situation or even that they're right.

This is just funny. The pros "don't know what they're talking about". Why do you think they are pros? Because they DO know what they are talking about. Many of them are hands on in development of the technology that is part of the digital process today.

You do notice that you've contradicted yourself here ? First you say maths does'nt matter and here,all of a sudden they're famous because they do know about maths.

I've seen Bob Katz claim that a properly dithered 16 bit recording can have 120 dB of dynamics. Which is kinda funny since the theoretical limit is 96.
And did you read when Roger Nichols made a complete ass of himself in SOS ? It was painful to read.

So if Trevor Horn (or whatever favourite producer you have. TH is one of mine) claims that changing the colours(or skin) of your host makes the sound better,that it is right ? Just because he sold more records than me ?

If he believe's that,it still won't prevent him from making great sounding records but it still does'nt make it right.

This buisness is hard enough without believing in woodoo.

A couple of questions:
1. Why are you so afraid to try for yourself ? It's so easy to prove you wrong,yet you still refuse to try.
2. What on earth does record sales has to do with if you're wrong or not ? I don't see the correlation.
3. Do you even know the difference between integer and floating point ?
4. What does gain staging on a fixed point host has to do with a floating point host ? You do realize they're 2 completley different matters ?
5. Have you ever heard of placebo ? I can't count how many times i've seen people telling that they changed the EQ on a channel that was'nt even on and heard a difference. I've even done it myself a few times. You swear there is a difference and then you look up and see you're adjusting the wrong eq.

You obviously can't hear those differences. Engineering is a tallent, and we do not all possess it.

You obviously can hear the difference when there isn't one.Engineering is a tallent, and we do not all possess it.

[music-tech]

Well, since this is quickly turning into a flame fest, with people somehow using what I said against each other, let me at least clarify my position.

1. I do believe/know that you need to watch your gain staging on the analog side of things, and that it will make a difference in the sound. This applies to both the A/D and the D/A side of things.

2. As far as mixing inside of the box goes, and whether or not lowering levels on a floating point system will have some big noticeable improvement on the sound... I do NOT have an opinion either way as I have not tested it myself. I did point out in my post a couple of things that could POSSIBLY have an effect on the sound, such as fixed point plug-ins within channels, or the remote possibility that the modulating nature of the digital noise floor with floating point systems could somehow be more exaggerated when pushing all the levels hotter -- but, that was just a really wild guess as to the only thing that I can think of that could cause an audible difference, assuming that the programmers did a proper job with the math on the mix engine.

3. Fixed point systems obviously can't be driven to clipping, unless you do it on purpose because you like that sound (I've purposely clipped kick and snare tracks on occasion to add a little extra bite to the attack). There are also stories all over the net from various pros that the earlier Pro Tools system, with its fixed point mixing engine, simply didn't sound good... lots of reports of that, in fact, which means either people were driving it into clipping distortion, or Digidesign simply screwed up the math. I think part of this current discussion is carrying over from those days when, apparently, it was true with Pro Tools that you needed to be careful of your mix levels inside the box. Again, I have no first hand experience of this.

Regarding the comment about Bob Katz saying that proper dithering can give you more than the theoretical dynamic range than 16 bits allows, this is actually true if you study up on dithering and how it works. Dithering basically allows you to hear signals that are below the noise floor and the lower limit of 16 bits since you are adding a low level shaped noise signal to the audio before you truncate to 16 bits... thus, some very low level signal that would have been lost if you did not dither before truncating to 16 bits, now can actually be represented by just 16 bits because you've added some fixed low level noise, and that addition brings that audio up to where it can be represented by 16 bits. Probably not a great explanation, but go study the articles on dither by mastering engineers who can explain it much better than myself.

I'll also say that it is my opinion that you should not believe everything someone says simply because they have some gold or platinum records. Many of these pros are not formally trained in any type of engineering degree, and learned on the job. Now days, with the internet, we all know how easy it is for myths to spread quickly and become accepted as the truth. Many pros, including myself, are guilty of that... although, after having been burned once by accepting something just because some big names said that, I'm much more careful to check out my facts first. I'm a full time professional, but not a well known name by any means. I also have a degree in Electrical and Computer Engineering, so I tend to side with the laws of math and known physics, rather than simply believing some wide spread myths. And, I know how easy it is to fool your ears if a test is not set up in a very scientific and controlled way. As mentioned, I've also tricked myself into thinking I was making changes with an EQ, only to find out it was bypassed.

Oops, gotta run! Client here

[Peel]

Well, this is kind of a minor point that passed by a while ago, but the statement about dithering increasing dynamic range is not actually wrong (at least not in the sense Katz is talking about). Dithering increases the dynamic range audible in a quantized signal, strange as it sounds, by reducing quantization distortion (which heavily masks the source signal at low levels) at the expense of raising background noise (which doesn't mask the original signal as much). This is something you can test yourself; it's like you're going beyond the theoretical 6dB-per-bit limit -- whether it is equivalent to 120dB is just an estimate.
Having said that, I really don't want to sound like an ass, but where02190, you've been consistently, absolutely, black & white, 100% wrong on this issue (among others) and the only reason you're not being told so each & every time you post is that few people have the patience to know where to start trying to make you understand that. And the majority opinion of established names in the field on this issue is also not remotely close to siding with you, no matter what you claim. I hate to be such a dick about a simple disagreement, especially since I know it won't change your mind, but it's not like you're always humble when you disagree with someone either.

[MacZero]

[bduffy]

Sorry - why the limiter on the main bus?