Recording at lower levels in a DAW?

[TechnoCowboy]

Hello. For some reason I cannot find the thread where somebody was talking about how it was important to record and work at lower levels while in the box. For instance keeping all your levels at -16ish , but I cannot remember why this was? He gave an explanation as to why the digital sound sounded better if kept at lower levels. can anyone please explain why or give me a link?

Thanks so much.
TC

[Sunbreak Music]

John Scrip has a great article on it here:

Proper Audio Recording Levels | Rants, Articles | My MASSIVE Blog

[TechnoCowboy]

thanks!!

[Riccardo]

Glad you found it.

Just as a reminder: -16 dBfs is not a low or lowish level!

[chester]

hi,

a digital system works with bit's!

1Bit is ~6dBfs [dBFullscale - Ends at 0dbfs -> it will clip!! digital distortions!]

24bit*6dbfs= 144dB !!!

that means if you record with -16dBfs you will loose like 2-3 bits inside the DAW, which in most cases are not critical. if you leave a headroom its better to work with equalizers when you boost them, cause you will have a headroom of ~ -16dBfs. so (not needed!! imho - if so you've done something wrong at source) you can boost like 16dBfs before it clips (distortion) inside your DAW.

And in some cases -16dBfs are associated with Bob Katz's monitoring plan. Means that your monitors are calibrated to ~83dB at -16dBfs with a white noise signal.

hope that helps.

bye ches

[Sunbreak Music]

I think it's really about gain staging and overloading your analog front end whilst trying to max your levels in your DAW to as close to 0dbfs as possible.

In other words, knowing the useable headroom of your pre and the headroom of your converters so the signal is optimized.

PS. Chester, you should read John's article.

[beyarecords]

You may find the following posts, by Paul Frindle, of interest:

2buss comparison: Fatso vs. Oxford Dynamics

Q for Paul Frindle

[Dean Roddey]

If you have decent converters, they should provide you with the metering you need to know what leves are hitting them, right? I cannot imagine that a reasonably good set of converters cannot handle a -6dB peak without distortion.

And, if you have a good pre-amp, it should also provide you with input and output levels, so you know if you are pushing it too hard.

Obviously you don't need to work like in the tape days and max out the levels. But peaking in the -12 to -6dB range (with the RMS some amount below that, depending on the type of content), I just can't see how that could be a problem if you have decent equipment.

It seems that, if you need boosts of 16dB, don't you already have far worse things to worry about?

And I've always found it interesting that when it comes to this, all the arguments about gain staging seem to be thrown out the window. Why isn't it bad gain staging to bring down all of your levels quite low, and in many cases pull down faders to get the other tracks a appropriate levels to match the lowest ones, to do all the processing with reduced resolution, then push the whole thing up 3 or 4 bits worth in the end? That's exactly the same as bad gain staging, it sounds to me.

I'm just pointng that out as an observation, not trying to make any particularly argument out of it. But it seems to me that if you track in the -12dB to -6dB peak range, with around a -20 to -16'ish RMS depending on material, that you'll end up with about -9dB'ish master bus levels (peak), which should be fine, it shouldn't cause any distortion in reasonably well made equipment, and you still have plenty of room to make reasonable mastering adjustments I would think. If you need more than that, perhaps a re-mix might be in order?

Anyway, just just being argumentative for the interest of a good discussion.

[chester]

hi Cass,

John's article is mostly what i am thinking of a good recording! Record with correct gainstages with a good headroom inside your DAW (also analog) and it will sound more open, clear etc. There is no need to record hot, squashed or whatever - in most cases it will sound thin instead of round, big, with balls - you name it.

in the age of 24Bit there is no need to record neaby the 0dbFs! leave yourself a good amount of headroom -> the converters will not overdrive, you will have a better dynamic range (without distortions) and also a better frequency response.

bye ches

Originally Posted by Sunbreak Music

I think it's really about gain staging and overloading your analog front end whilst trying to max your levels in your DAW to as close to 0dbfs as possible.

In other words, knowing the useable headroom of your pre and the headroom of your converters so the signal is optimized.

PS. Chester, you should read John's article.

[Bob Yordan]

+1

I try to avoid getting the main stereo output touching red while
recording and mixing music within C4. Leaving headroom for the
mastering. While mixing I usually add my CL1MK2 on the master bus
(to glue) but it is never used for squasching the mix and my SLAM on a group
(processing drums & keyboards).

best wishes

[Franz]

Ok.

But if you keep your levels down in the -18 dBfs-region throughout recording and mixing, and you are mastering the material yourself, not using outboards. How are you supposed to work your master up to 0 dBfs without artefacts from makeup-gains and normalizing-functions?

[peeder]

You should gain stage everything for the best SOUND overall, not worry about meters. Digital gain is the very least of worries.

Insisting that a specific tracking or mixing level is optimal is like insisting that a specific guitar is optimal.

Nope. tutt

Why ask about tracking or mixing in a mastering forum?

[Lagerfeldt]

Originally Posted by Dean Roddey

Why isn't it bad gain staging to bring down all of your levels quite low, and in many cases pull down faders to get the other tracks a appropriate levels to match the lowest ones, to do all the processing with reduced resolution, then push the whole thing up 3 or 4 bits worth in the end?

Floating point engine.

[Audio Hombre]

Originally Posted by chester

hi Cass,

John's article is mostly what i am thinking of a good recording! Record with correct gainstages with a good headroom inside your DAW (also analog) and it will sound more open, clear etc. There is no need to record hot, squashed or whatever - in most cases it will sound thin instead of round, big, with balls - you name it.

in the age of 24Bit there is no need to record neaby the 0dbFs! leave yourself a good amount of headroom -> the converters will not overdrive, you will have a better dynamic range (without distortions) and also a better frequency response.

bye ches

+1 i agree and humbly disagree with sunbreak. i never,ever, try and "max your levels in your DAW to as close to 0dbfs as possible."

[MASSIVE Master]

Originally Posted by Franz

Ok.

But if you keep your levels down in the -18 dBfs-region throughout recording and mixing, and you are mastering the material yourself, not using outboards. How are you supposed to work your master up to 0 dBfs without artefacts from makeup-gains and normalizing-functions?

I would *SO* much rather add gobs of digital gain to a "softer" CLEAR signal than add a little bit of ANY gain to a signal that was overdriven on the front end and then mistreated through every following step...

Normalizing functions...?

[Sunbreak Music]

Originally Posted by Audio Hombre

+1 i agree and humbly disagree with sunbreak. i never,ever, try and "max your levels in your DAW to as close to 0dbfs as possible."

You shouldn't humbly disagree--you should strongly disagree!!!!

I wasn't advocating that--the opposite in fact. I was speaking to what John's article addresses, as opposed to things like intersample clipping and other reasons not to record as hot.

These discussions quickly splinter into several directions--and at a glance my post was confusing.

Good on ya'!

[Dean Roddey]

Floating point engine.

That doesn't really make any difference, AFAIK. A floating point engine obviously has some advantages, but if you lower all the levels of all tracks, process them at that lower level, then push them back up, that's exactly what bad gain staging is.

If a floating point engine made gain staging a non-issue, you could turn down the sensitivity on your converters and store everything at like -50dB (analog input chain gain staging would be done as normal) and do the whole mix and push it up 50dB at the end and have it sound just as good as if you'd stored it at more normal levels. I don't think that's the case, right? All the processing introduces noise, and the more bits they have to work with, the more the signal stays up above that noise.

Seems to make sense to me anyway.

[Lagerfeldt]

Originally Posted by Dean Roddey

That doesn't really make any difference, AFAIK. A floating point engine obviously has some advantages, but if you lower all the levels of all tracks, process them at that lower level, then push them back up, that's exactly what bad gain staging is.

That's not how it works, if we're talking inside a DAW. Maybe you're only referring to if the actual recordings were low?

As for inside a DAW, the audio isn't processed at the lower level shown by the fader, it's using the actual recorded level of the 24 bit audio file for 32 bit or 64 bit floating point internal processing in the plug-in. This then is either output from the plug-in as a 32 bit floating point signal or a truncated (or dithered) 24 bit signal to the summing engine.

Turn down the levels on the individual channel faders and turn the master up - or turn up the individual channels and turn the master down: makes no difference to the final outcome of the mix. The two files will null completely in a floating point engine.

So there's absolutely no signal loss by lowering your fades in a DAW like this. On the contrary it will have the benefit of not overloading the input side of the busses and master output, very relevant assuming you have plug-ins on at least your busses. It's fairly easy to test yourself using Waves plug-ins which do not have input attenuation.

However, don't come too close to 0 dBFS either when recording or processing in plug-ins or you might get in trouble because you can exceed internal headroom (even when exclusively cutting in an EQ, due to phase changes) or get intersample peaks. Also some A/D converters are not very happy about signals close to full scale (RME ADI-8 DS for instance). Keep below -3 dBFS. Haven't tested this on the new QS model which might be better.

[Sugarnutz]

Here's a massive thread over at Terry Manning's forum which covers the subject pretty well: PSW Recording Forums: Whatever Works => Digital tracking with low levels = better...is this new???.

Most convertors are calibrated where a +4 level from say a mic pre reads at -18 to -15 (depends on the manufacturer) in the DAW. This is called "Headroom", which is really nice thing to have when working in digital, of course YMMV as usual.

[peeder]

Most modern music benefits from distortion.

Some distortion is better than others.

Distortion often means exceeding headroom. In the right place, with the right signal, with the right processor. Whether hardware or plugin.

Yes plugins can have saturation curves too. Some of them quite rewarding. So can converters.

Avoiding distortion at all times may earn you a pat on the head from a few mastering engineers, but it is not going to bag you a grammy. Sorry.

[Old Goat]

Thanks, John Scrip, for my 'ah hah' moment du jour. I have been operating under the "hot as you can" mantra. I see that some of my mixing and mastering problems are probably self-inflicted...

[Sunbreak Music]

Originally Posted by peeder

Most modern music benefits from distortion.

Some distortion is better than others.

Avoiding distortion at all times may earn you a pat on the head from a few mastering engineers, but it is not going to bag you a grammy. Sorry.

John's article is really an educational piece, geared toward those who might not understand proper gain staging into their daws.

If someone doesn't understand the ideas of calibration, headroom, and gain staging, odds are they aren't heading to Grammy-ville, so I don't quite get your point (other than a swipe at a few who are trying to do some education and explain some concepts).

I suppose we're pretty OT now.

[peeder]

Originally Posted by Sunbreak Music

John's article is really an educational piece, geared toward those who might not understand proper gain staging into their daws.

If someone doesn't understand the ideas of calibration, headroom, and gain staging, odds are they aren't heading to Grammy-ville, so I don't quite get your point, other than a swipe at a few who are trying to do some education and explain some concepts.

I suppose we're pretty OT now.

My point is that enough people screeching that you have to keep tracking and mixing levels low at all costs is 100% exactly as stupid as the people screeching that you have to hit converters as hot as you can at all times.

I've already seen on this board people suggesting the addition of attenuators to a chain to bring a signal that was peaking at an ADC at -6dbFS (no distortion likely from the ADC) to -18dbFS because of just this hysteria. And that analog attenuator is going to damage the signal a lot more than printing at -6dbFS would. So would not driving the chain hot enough to get off the noise floor prior to the ADC.

Not to mention the opportunity cost of missing out on some glorious THD from your analog gear.

Yes we got it you don't have to hit ADCs as hot as you can, and yes we got it some plugins can saturate and clip as they near and exceed 0dbR.

Everyone who is intelligent should tell both sides of the story, and since no one else is doing so, here I am. Very much on topic.

[Adam Dempsey]

Originally Posted by MASSIVE Master

I would *SO* much rather add gobs of digital gain to a "softer" CLEAR signal than add a little bit of ANY gain to a signal that was overdriven on the front end and then mistreated through every following step...

Well said, John.
Only further to that... good, clean, analogue gain. One example: I love the API 550m's added 1.5dB gain when just running through them, or a touch of the LTD-2's gain when called for. A lower, cleaner mix (edit*: than one hitting full scale too often or subjected to earlier non-euphonic and quite possibly previously unheard distortions) gives more scope in analogue gain staging, despite the fact we can knock the mix level down on the console's input if needed. thumbsup

[MASSIVE Master]

Some distortion is better than others.

And I wouldn't argue that -- But I can add it, fake it, fabricate it, etc., etc. -- But I can't take it away if it's already there.

And no doubt, it's used creatively all the time -- Just ask any Marshall amplifier. The approach I'm getting at is that a lot of gear -- especially the "budget friendly" stuff that a lot of people are using, just turns to crap quickly and easily when pushed - while many of them sound quite decent if levels are a little more conservative. "Conservative" meaning "the same signal you'd probably be using if you were hitting tape" - as opposed to the signals that many people think of as "normal" that would cook right through analog tape.

And even then - It might work well on this and that -- I had a preamp that started to break up readily and audibly whenever it went over around 0dBVU. I'd run bass through it, some vocalists, etc. But not everything...

Personally, I wish people didn't consider it "lower" levels... I've always (even back when 16-bit was the only game in Digitalville) thought of it as "normal" levels. The front end is designed to work at a particular level -- It doesn't care whether the recording medium is analog or digital - That level is the same.

[peeder]

Originally Posted by MASSIVE Master

And I wouldn't argue that -- But I can add it, fake it, fabricate it, etc., etc. -- But I can't take it away if it's already there.

And no doubt, it's used creatively all the time -- Just ask any Marshall amplifier. The approach I'm getting at is that a lot of gear -- especially the "budget friendly" stuff that a lot of people are using, just turns to crap quickly and easily when pushed - while many of them sound quite decent if levels are a little more conservative. "Conservative" meaning "the same signal you'd probably be using if you were hitting tape" - as opposed to the signals that many people think of as "normal" that would cook right through analog tape.

And even then - It might work well on this and that -- I had a preamp that started to break up readily and audibly whenever it went over around 0dBVU. I'd run bass through it, some vocalists, etc. But not everything...

Personally, I wish people didn't consider it "lower" levels... I've always (even back when 16-bit was the only game in Digitalville) thought of it as "normal" levels. The front end is designed to work at a particular level -- It doesn't care whether the recording medium is analog or digital - That level is the same.

I can't get back the opportunity to distort signals in specific ways once the gain structure has been altered. For instance, let's say I track from a preamp to a compressor to an ADC. I can't post-facto go back out through an attenuator and back through the chain to distort it the way I might like to, because the signal has now been compressed and it will not distort at the same points. And my preamp or compressor might have great output transformers that I'd like to drive.

More importantly, the musician will not be able to interact with the distortion done after the fact. Imagine Hendrix playing a solo with a clean DI sound rather than milking feedback out of a screaming Orange or Plexi. Same goes true with a vocalist driving an LA-2A.

The answer to all of this is to feel free with your levels, and let your ears guide you.

Digital gain/trim is perfect enough you don't have to worry about using it, and it's very easy to get your track levels right across the mix with the input trim on your first plugin or a trim plugin. It's very easy to pull down a master fader on a bus if you're pushing the bus too hot.

The problem isn't so much blind adherence to philosophies of hitting an ADC hot or maintaining nominal levels. We should avoid either of those tarpits with equal vigor. The problem is people are not in a habit of listening carefully enough to what they are doing at each stage, and understanding how each gain stage fits into the picture.

[Bob Olhsson]

Nobody can predict the actual levels coming from a microphone during the perfect take. There will always be an error.

Many of us have had the experience that it is far better to error on the low side than on the high side with most common digital converters and we simply offer this as advice and not dogma.

[Dean Roddey]

That's not how it works, if we're talking inside a DAW. Maybe you're only referring to if the actual recordings were low?

Well, yeh, partly I was. The issue is, as was just ranted on, of having to record very low to avoid distrortion. I was saying that, if the argument is that gain staging is irrelevant here (record everything low, process it at those low levels, i.e. fewer bits for the plugs to work on), then push it all up at the end, then why does it matter anywhere once it's inside the DAW?

I think that everyone would argue that it's bad gain staging to have plugin one pull the level down by 18dB and plugin two push it back up by 18dB, right? From the perspective not of the summing engine, but of the noise introduced by plugins, that's a bad thing still, isn't it?

If so, then I don't see why people would argue to record at really low levels, run it throguh all the plugs at those low levels, and then push it back up at the end. It's exactly the same as bad gain staging.

I understand the arguments, that you analog chain and converters won't be pushed too hard. And if that was a really bad problem, obviously it would be worth doing in a 24 bit system. But if you have metering on your analog equipment and you have it on your converters, then you know what the levels are all the way through and whether you are pushing them too hard or not. If yo have good equipment, as long as you keep the peaks below -6dB'ish so that your final master bus levels leave you room to work at the end, and you know that you aren't pushing anything in the chain too hard, I'm not sure I see what all the fuss is about and why it's not just bad gain staging for nothing, or sub-optimal gain staging anyway.

Obviously, as mentioned above, if you are dealing with a very inconsistent level and aren't using a limiter or something, you have to be careful. But if it's not something practical like that, I'm not sure I see the point as long as you have the good toys and you know you aren't pushing them beyond their specs.

[peeder]

Originally Posted by Dean Roddey

I think that everyone would argue that it's bad gain staging to have plugin one pull the level down by 18dB and plugin two push it back up by 18dB, right? From the perspective not of the summing engine, but of the noise introduced by plugins, that's a bad thing still, isn't it?

In the modern floating-point digital realm, noise and distortion basically don't exist unless a programer takes pains to create them. In 32-bit float, you have 900db of headroom and 150db of dynamic range (no, those aren't typos...digital is odd that way). Clean gain and trim is ubiquitous and also just about unavoidable.

Virtually no plugins introduce noise...some tape sims do, and of course signal generators.

You can overdrive a bus in a floating point DAW by the mentioned 900db, bring the signal back down 1400db, pull it back up 500db and null with what you started with to a full 24bits (144db) of precision. You can do that through some plugins too.

Some plugin makers want plugins to clip at 0dbR (an arbitrary point where the DAW maker flashes a red clip light, not to be confused with the clip lights they flash that aren't arbitrary at a converter or the file system), and if you have one of those in your chain, then it establishes that as an upper limit for consequence-free level. The plugin maker has to purposefully add this clipping behavior and any other saturation behavior; frankly I have no idea why many of them do when it's just going square at 0dbR and I really wish they wouldn't (it may be for legacy compatibility with old fixed point systems like TDM that they did this, feh I say to that ).

[FBM]

Hi, Some VST's produce noise because they use samples (with noise in the recordings) also in a 32bits floatingpoint DAW and some clip at 0db, because they work with 24bits in the 32bits environment . A good example is TC powercore they work with 24 bits!
Dvid