DAW Tracking Levels too hot?? Looking for guidance & resources

[alexdingley]

I've been listening to a bunch of podcasts lately & doing some reading on the subject of improving my mixes — and one of the themes I keep hearing / reading about is, essentially "recordists who track things too hot in the DAW are falling prey to a misguided goal... those fat / beefy waveforms look good on the screen, but you'll get better results if you track quieter." / "you don't have to get levels that are just shy of clipping for the audio to be 'full-resolution' anymore."

... as I hear this and read this, I'm thinking "uh oh... dat me!" — I've definitely fallen into that habit of trying to "capture all the bits" — and what I'm repeatedly hearing is that this isn't needed and can possibly prevent me from getting a good-sounding end result. Trouble-is... I'm having difficulty finding really good guidelines on "how low is low enough, but not too low... when tracking with my DAW?" Also, I'm wondering if there are any tell-tale signs that my DAW levels are tracked too hot?

• Do you find that each brand / interface has a sort of sweet spot input-level wise... and if so, how do I go about uncovering mine? (other than track many many many things and learn it over time.... or hey! if that's the only way, then lemme know.)

• Do you find that it's more of a source-dependent thing? (Track drums with a little more/ little less headroom, but track vocals hotter??)

• Any good video tutorials / explanations / on-line resources / those old paper-thingies... with the paper glued together at one side and all the words type into them ??

[A.David.MacKinnon]

I aim for average levels around -18. On Pro Tools meters that's right about where the yellow band starts.
If your tracks are at unity and the master bus is getting slammed then your tracks are too hot.

[MoreSpaceEcho]

there's def no good reason to be tracking stuff right up near 0dbfs. if you're recording at 24 bit (and why wouldn't you), you have more than enough bits to capture everything you need to capture and still leave plenty of headroom at the top.

there's no hard and fast rules, and this is not something that needs overthinking, just aim for peaks around -12dbfs, average levels in the -20s and you will be fine.

[Nick Sevilla]

(Edited down for facetiousness / clarity)

I'm having difficulty finding really good guidelines on "how low is low enough, but not too low... when tracking with my DAW?"

Also, I'm wondering if there are any tell-tale signs that my DAW levels are tracked too hot?

• Do you find that each brand / interface has a sort of sweet spot input-level wise... and if so, how do I go about uncovering mine? (other than track many many many things and learn it over time.... or hey! if that's the only way, then lemme know.)

• Do you find that it's more of a source-dependent thing? (Track drums with a little more/ little less headroom, but track vocals hotter??)

• Any good video tutorials / explanations / on-line resources / those old paper-thingies... with the paper glued together at one side and all the words type into them ??

Hi Alex,

As a general rule of thumb, you want to follow a practice where your PEAK levels (the transients, which are the loudest parts of a sound) are hitting well below a reasonable dBFS level, such as, -14dBFS, for example. I usually stay under this level and generally have had good success with avoiding problems, except for HIGH TRANSIENT sources, where I may go as low as -24 dBFS, depending on the source.

Tell tale signs can be small distortion artifacts in the Transients. To make a test for yourself, use an extremely high transient level source, such as a metal percussion triangle hit with a metal rod, or a cowbell, again hit with a metal object. What you are looking for is strange shape at the highest peak, like a sudden change in the waveform before hitting the top peak. This may indicate that something in your converter is getting distorted. Make sure to not have the microphone LESS than 3 feet to 5 feet from this type of source. You could distort the diaphragm or the internal mic electronics. Also, do not turn up your mic preamplifier too much. Make it less than you normally would. Do not worry about "noise" at this point, as this is not what you'll be worried about and measuring. To test your converter, try to hit the PEAKS on its meters at approximately -18 dBFS maximum. Hit the instrument several times, vary your intensity a little bit, play naturally. Record about 20-30 seconds of a simple beat. try hitting a little harder and a little softer, but be consistent as far as how far from the mic you are, don't move around.

You do need to understand that LED and virtual metering IS NOT ACCURATE. It is an approximation. The only meterings that are accurate cannot be used for real time tracking, because the converters themselves are also not perfect in their capturing of the audio, and subsequent conversion to an electrical signal, which happens BEFORE it becomes a digital signal. There are a few plug ins which can detect inter sample peaks, but sadly only after the fact. Which is ok for this test, if you have one of those plug ins. Also, digital meters usually cannot display INTER SAMPLE PEAKS, which are peaks which occur between samples. So, I suggest you do this test at 96kHz, or higher, if possible. Otherwise, use your most common sample rate, the one you usually use for most of your projects.

ADC works like this:
You have a real instrument >
Microphone (possible distortion can happen here) >
Mic preamplifier (more distortion possible here) >
Analog stage of converter (more distortion can happen here) >
Analog conversion into a voltage signal (more distortion can happen here) >
Digital converter chip (more distortion can happen here) >
Digital audio stream that goes onto a hard disk drive.

So, you have a whopping 5 possible places where distortion can happen. Now, today's converters typically are set up internally so it is VERY HARD to have distortion happen before the signal turns into zeroes and ones. But, it can and does happen with semi pro units, and some older professional ones too. Thus my suggestion that you never EVER hit above -14 dBFS (or even lower, depending on your converter and the test you did with the triangle). Things like amplified electric guitars , where the amplifier cuts off transients already, you can record them as high as -10 dBFS. But percussions, drums, and instruments that you have NEVER recorded before, stay below -18dBFS, or even -24 dBFS. Remember, you can always turn them up after the fact digitally, and lose nearly no quality at all, but you cannot undo these tiny distortions though.

After recording your triangle or other high transient percussion instrument, look at the waveform as zoomed in as possible. Look at the different waveforms at different hit levels. Are they the same? At what point do they start looking different? If you do this right, you can find where your converter is distorting the signal. Remember to have your microphone a MINIMUM 3-5 feet from the source, or further if need be, and also your mic pre amp turned lower than normal.

Helpful hints:

1. Microphone diaphragm distortions look like sudden square waves, and depending on the type and size of the diaphragm, may have ripples after the diaphragm has hit its distortion point.
2. Microphone internal circuit distortion looks like an overdrive guitar pedal, or in extreme cases, like a heavy metal pedal: the waveform peaks are rounded off both at the top and bottom to varying degrees, also with a slow or fast rippling or disappearance of this distortion as the waveform level drops off.
3. Microphone preamplifiers, especially Neve like class A or A/B, are also like distortion pedals, they cause the peak of waveforms to curve up. If you drive a Neve preamp, or any solid state preamp TOO HARD though, and you will see a ripple effect afterwards, which is the op amps returning to normal voltages. Fun looking waves, those! Tube preamplifiers will distort consistently above a certain input level, and looks a lot like tube guitar amps: They turn the signal into a soft looking turd, and as the level comes down, they slowly clean up the sound.
4. Analog converter stages also will distort the peaks into curvy shapes, but are a LOT less forgiving than the previous stages. these also will occur at the SAME level every time.
5. Digital chips COMPLETELY FUCK UP the signal. You get a sort of square wave "all or nothing" type distortion.

Hopefully this serves as a tutorial of sorts.

Good luck finding your converter's distortion point!!!

[vvv]

[alexdingley]

Edited for brevity ...So, you have a whopping 5 possible places where distortion can happen. Now, today's converters typically are set up internally so it is VERY HARD to have distortion happen before the signal turns into zeroes and ones. But, it can and does happen with semi pro units, and some older professional ones too. Thus my suggestion that you never EVER hit above -14 dBFS (or even lower, depending on your converter and the test you did with the triangle). Things like amplified electric guitars , where the amplifier cuts off transients already, you can record them as high as -10 dBFS. But percussions, drums, and instruments that you have NEVER recorded before, stay below -18dBFS, or even -24 dBFS. Remember, you can always turn them up after the fact digitally, and lose nearly no quality at all, but you cannot undo these tiny distortions though...

...You do need to understand that LED and virtual metering IS NOT ACCURATE. It is an approximation. The only meterings that are accurate cannot be used for real time tracking, because the converters themselves are also not perfect in their capturing of the audio, and subsequent conversion to an electrical signal, which happens BEFORE it becomes a digital signal. There are a few plug ins which can detect inter sample peaks, but sadly only after the fact. Which is ok for this test, if you have one of those plug ins. Also, digital meters usually cannot display INTER SAMPLE PEAKS, which are peaks which occur between samples. So, I suggest you do this test at 96kHz, or higher, if possible. Otherwise, use your most common sample rate, the one you usually use for most of your projects...

...Hopefully this serves as a tutorial of sorts. Good luck finding your converter's distortion point!!!

Okay — woah... that's super helpful Nick! PS... thanks for always being a willing resource on this board!

Quick questions: Are you suggesting that the DAW meters are no good for metering on the way in... but are good for metering what's been recorded? And are the DAW virtual meters, on input, so inaccurate that I should not rely on them to tell if I'm staying within safe limits.... ? or are you saying that they're just not accurate enough to tell the difference between "-18" and "-17" on the way in... so that they're less-than-ideal for any calibration steps?

[alexdingley]

follow up question: is the primary reason to record things lower (at which point, my peaks will be in the -18DBBF to -24DBFS range) solely to prevent digital distortions, or does having more headroom in the track also cause the signal to process "better" in plug-ins that get used, or at the point of summing digitally?

I can see how tracking things with lower input levels is going to help avoid that issue I used to hear people describe pretty often... "I tracked a bunch of stuff with great looking waveforms, but when my faders are anywhere near unity, my ITB mix is clipping like crazy and way in the red... so I have to pull all my DAW faders waaaaay down to make a mix that works."

...side note: I just realized that nowhere in my old "textbooks" did this particular topic get covered... It's funny that I went to recording school 20yrs ago and never learned this (seemingly crucial) best-practice. I learned a hell of a lot about the principles of recording on tape & into digital... the basics of gain-staging are covered, but the fine art of "getting good levels" doesn't really show up much in any of my old texts... Granted, PT was really just coming into popularity then, but still... was the whole "fat waves aren't always good waves" thing a more recent revelation to everyone??? Or have I just been missing this (very important) boat, the whole time?

[A.David.MacKinnon]

If your text books are from the era of 16 bit recording then you’d be advised to record as hot as possible without clipping to get the best resolution. It's just not an issue anymore at 24 bit.
Beyond that, the when anolog end of your gear is built to function best hitting 0 on the vu meter. That's about -18 in digital

[Nick Sevilla]

Okay — woah... that's super helpful Nick! PS... thanks for always being a willing resource on this board!

Quick questions: Are you suggesting that the DAW meters are no good for metering on the way in... but are good for metering what's been recorded? And are the DAW virtual meters, on input, so inaccurate that I should not rely on them to tell if I'm staying within safe limits.... ? or are you saying that they're just not accurate enough to tell the difference between "-18" and "-17" on the way in... so that they're less-than-ideal for any calibration steps?

Hi Alex,

All I am suggesting is that digital meters (the ones in the converter itself) are still not fast enough for certain transients. So you should not rely on them when you are recording percussive instruments and things like that. Like ANY metering, they are "close enough" to give you an idea of where your levels are. BUT you should always do a test recording, and listening to that and looking at the waveform before doing any final recordings for actual use in production. Always.

The DAW similarly, at least the ones that are the default ones that come with your DAW, are also only good to get an IDEA of where your levels are.

If you really REALLY need to get "meter anal", you can get something like this plug in:

https://www.waves.com/plugins/wlm-loudn ... tart-guide

This plug in, can only really give you accuracy on an already recorded file, or on your mix buss. Again, when recording, it is not fast enough to "catch" really fast, loud transients. These transients are so fast, there really isn't anything on the market that can meter them properly. We are talking a few uS second lengths here (1 pico second = 0.000000000001 second, or one billionth of a second), for the actual peak itself. Tiny stuff!!!

Calibration is, as I suggested, using slower than these speeds metering, and analyzing the actual waveform to CONFIRM where your equipment starts distorting these transients, a decent way to avoid distortion on those types of sound sources. You use the meters to "rough in" your levels, do a test recording, and then analyze the resulting audio file to "see" where distortion, if any, has occurred. You can use, in Pro Tools, for example, the "gain" plug in, and "analyze" the actual dBFS level where your peak actually is, rather than use a metering plug in. Then you KNOW at what dBFS you are nice and clean, and at what dBFS you are getting peak distortions.

I suggest you just go ahead and do a little test recordings, and learn where in your signal chain things get distorted at each device.

Cheers

[Nick Sevilla]

I can see how tracking things with lower input levels is going to help avoid that issue I used to hear people describe pretty often... "I tracked a bunch of stuff with great looking waveforms, but when my faders are anywhere near unity, my ITB mix is clipping like crazy and way in the red... so I have to pull all my DAW faders waaaaay down to make a mix that works."

...side note: I just realized that nowhere in my old "textbooks" did this particular topic get covered... It's funny that I went to recording school 20yrs ago and never learned this (seemingly crucial) best-practice. I learned a hell of a lot about the principles of recording on tape & into digital... the basics of gain-staging are covered, but the fine art of "getting good levels" doesn't really show up much in any of my old texts... Granted, PT was really just coming into popularity then, but still... was the whole "fat waves aren't always good waves" thing a more recent revelation to everyone??? Or have I just been missing this (very important) boat, the whole time?

Here is a good analogy (hopefully) :

Your DAW mix buss is NOT a glass of water that can overflow in a nice sounding way, like an analog tape machine does.
No, your DAW mix buss is a WALL OF DEATH. If you ever hit that WALL OF DEATH, your mix DIES there, in a horribly distorting way.

Remember that to add decibels of any kind, you need HEADROOM. LOTS OF IT. So that when you have mixed in EVERY track for your song, the resulting mix levels NEVER get close to that WALL OF DEATH. I always mix with the peaks at -6 dBFS Maximum. And this, while using that Waves meter I linked to before. I usually put that in the mix track, AFTER the mix buss and everything else. It looks at the final mix audio.

I always am test printing when doing a final mix. And analyzing the hell out of that final audio mix file. You'd be surprised how much distortion can get through to your print, especially if you are not careful.

Helpful tip: Just like on analog equipment, you can and should set your mix levels properly. For example, your lead vocal level at the mix output, should approximately start out hitting around -12 dBFS. this is at the LOUDEST vocal parts. This gives you 12 dBFS headroom to being mixing with, as you add more instruments. Depending on how many instruments you have for a given song, your lead vocal should be there OR LOWER in the case of lots of tracks, so that as you mix around the all important lead vocal, you never hit that WALL OF DEATH.

And if you DO, take all your tracks, group them together (Pro Tools has an "all" group just for this BTW) and BRING THEM DOWN until you again are hitting that -6dBFS for the entire mix (all the tracks needed playing together, at the loudest part of the song).

Hope this helps. This BTW took me YEARS to understand. And it was never in any Goddamned book. Ever. Maybe I'll write one someday. Or keep a few secrets (wink wink).

[Nick Sevilla]

Here is another useful metering plug in:

https://www.waves.com/plugins/paz-analyzer

And LOL to the "fat wave" thing. Our saying is "If it looks like a turd, it sounds like one too". Avoid the crap.

Today's converters are ALL 24 bits. All of them. this means you have a whopping 124 to 144 dB of DYNAMIC RANGE.

Dynamic Range is the SIZE of Loud to Soft that a converter has. In easy terms, you can record a sound source that is 100% quiet all the way to 143 dB of loudness,without distortion. The pain threshold for the human ear begins at about 125 dB. So you can have another 19 dB of fun and games AFTER you are in terrible ear pain. Wonderful!!!

So, you NEVER EVER EVER NEVER EVER need to record ANYTHING above -10dBFS ever again. Those days are OVER.

Here, a nice chart:

http://www.hearnet.com/at_risk/risk_trivia.shtml

[Nick Sevilla]

Finally:

The louder the transient of a sound source, the LOWER the level you should be recording it at.

Example, that triangle percussion. That is undoubtedly one of the highest level transients we have in the studio normally.

Now, if you have a noisy synthesizer, of course, as long as it is things like pads and other non fast transient sounds, you can and should record them at a louder level, because then you can and should turn these DOWN in your mix. And guess what happens? You also are pushing DOWN their hissy noise crap in the mix. A common sense noise reduction technique without needing silly noise reduction plug ins.

I have an old ass Roland synth from the 1980s, and it's noise is at -60 dBFS. SO I always record it louder than needed, at about -14 dBFS, and then when I am mixing that in, usually the fader for that track may be at -23dBFS, a nice 9 dBs lower. Meaning I never get to hear that noisy hiss from that synth in the mix, because that noise is now at about -83 dBFS now. Lower than the original real synth of -60 dBFS.

[A.David.MacKinnon]

Example, that triangle percussion. That is undoubtedly one of the highest level transients we have in the studio normally.

I used to regularly play with a vibraphonist who also played glock with brass mallets. It was the loudest fucking thing I’ve ever heard. I couldnt be in the live room with him for more then 2 mins. When we’d play live the sound tech would always ask how he should mic the vibes/mallet percussion. We’d say put a mic anywhere and it’ll come through. He was always the loudest guy on stage.

[vvv]

FWIW, I find that if I start with the track I want to be the focus, ex., the lead vocal, peaking around -15 and build my mix from there (snare is often -14 to -16, kick is often -17 to -20, or a drum submix might be -10 to -14, guitars around -22 to -24, compressed/limited bass -24, etc.,) my mixes end up peaking at about -9 or so where I have bass, drums, vocals, two -3 guitars, mebbe a key pad ...

Obviously, the EQ of various tracks, if the guitars or bass are distorted, how compressed the drums are or even the vox, how many tracks of whatever are variables; I'm just trying to say what my limited little experience is recording rock, funk and alt.cuntry.

Nick, on that chart you linked it says, "The incidence of hearing loss in classical musicians has been estimated at 4-43%, in rock musicians 13-30%." Can you interpret that? I mean, is it that imprecise that they don't know where rock muso hearing loss is other than between 13 to 30% of us? Hang on, that phone won't stop ringing ...

[Nick Sevilla]

Nick, on that chart you linked it says, "The incidence of hearing loss in classical musicians has been estimated at 4-43%, in rock musicians 13-30%." Can you interpret that? I mean, is it that imprecise that they don't know where rock muso hearing loss is other than between 13 to 30% of us? Hang on, that phone won't stop ringing ...

LOL I think they have it backwards... at least all the rockers I have worked with always SAY WHAT?!? When I first talk to them.

As to classical musicians, they all do get a little hearing loss, but nowhere nearly as severe as the rockers.

The most common hearing loss among rockers is in one ear, the one NOT facing the crowd during their concerts, but facing their PA / monitors.
Thank technology in that we now have in ear monitoring, which means less volume going into the ear canal, and some protection from outside noises.

Cheers