Proper Audio Recording Levels
You're probably tracking too hot. Here's why...
This entry was viewed over 110,000 times in 2008, nearly 300,000 times in 2009, picked up by and published by Pro Sound Web in 2011 -- And is *still* waiting for tweaking... And I feel a bit guilty about it because (as of mid-2016), the e-mails just keep coming in. That said, evidently, the point is getting through. Occasionally, I might “bump” or “TTT” (To The Top) this article just to avoid scrolling.
This entry was viewed over 110,000 times in 2008, nearly 300,000 times in 2009, picked up by and published by Pro Sound Web in 2011 -- And is *still* waiting for tweaking... And I feel a bit guilty about it because (as of mid-2016), the e-mails just keep coming in. That said, evidently, the point is getting through. Occasionally, I might “bump” or “TTT” (To The Top) this article just to avoid scrolling.
NOTICE: If you don't want to read any this or just don't care to understand it, there's a "dumbed down" version at the bottom.
Let me get something out of the way here - I'm going to try to keep this very "fool proof" - I'm not trying to sound or present this very scientifically - This is just the rantings of hundreds and hundreds of posts on a dozen or more audio forums exploding like a volcano recorded with lots of headroom. I just hope to instill a basic understanding of why certain trends and common beliefs are just plain bad. And by the time you're done reading, and perhaps doing a little experimentation based on this, you won't need me to prove it. You'll know it yourself.
Is this a "miracle cure" for bad recordings? Normally, I'd say no. But with the dozens and dozens - Easily now into hundreds of e-mails, phone calls, letters, forum posts and other forms of communication from one after the next about how this advice has completely changed the way they view recording, I figured I should put it down in a centralized location. The sad part, is this should be common sense. To anyone that grew up "on tape" it probably is. To those brought up in 1's and 0's, it might not be so obvious.
So, if you've been struggling with recordings that sound "weak" or "small" or too dense or "just not 'pro' enough" then please, read on. If this is about you, you might think differently soon.
As a mastering engineer, I work on recordings from pretty much every level of experience. A few years ago, I noticed something unusual.
"Ultra rookie" recordings - Those made by people with little or no experience, sounded fine. They didn't know any better, so they didn't have enough rope to hang themselves with. "Pro" recordings sounded fine. They know what they're doing and/or are using gear with obscene amounts of usable headroom (explained later). The "middle of the road" engineers with a year or two - or much, much more experience -- Those are the recordings that sounded "small" and spectrally challenged. So after quizzing these people over months and months, I came up with the following conclusion...
You're probably recording too hot.
And it's absolutely ruining recording after recording after recording. And it's the simplest thing in the universe to correct.
I know, I know - "It says in the manual to record as hot as you can without clipping." Well, I'm going to flat-out call that B.S. and I'm going to back it up with a simple (if not somewhat time-consuming) experiment.
Also as a mastering engineer, let's get something straight -- I don't like the "loudness war" going on - But I'm as guilty as the next in contributing to it. I can't fight it, as much as I try. Hopefully it'll be over some day. HOWEVER - With the quest for LOUD, there are a LOT of engineers out there shooting themselves in the foot before they even know how to aim. They think that tracking loud and mixing loud contributes to a louder recording after the mastering phase. THIS IS ABSOLUTELY UNTRUE and it's generally the best way to make sure that your recording will NOT have the "loudness potential" of the average commercial release.
CLEAN recordings - Recordings made with low distortion and good spectral balance - THOSE are the ones that handle the "abuse" of the mastering phase with flying colors.
This article isn't intended to give you some secret way of making louder recordings. But it will almost undoubtedly give you the ammo needed to make BETTER recordings. And those BETTER and CLEANER recordings are the ones that can be LOUDER recordings in the end.
First, let's get through a little nomenclature -
dBFS: Deci-Bel (one tenth of a Bel) Full Scale -- On the digital recording scale, -0dBFS is the hottest signal you can have. "All ones." Top of the scale, can't get hotter, etc. Always "minus" as you can never go higher - So the reading will always be a specified amount below 0.
Line Level / 0dBVU: Just what it says. Line level. 0dBVU on an analog VU (volume unit) meter. Pro (+4dBu) or consumer (-10dBv) level, it's line level. We can also refer to this as RMS (Root Mean, Squared), or a level over a specific amount of time. You *can* go above or below 0dBVU. It's simply a nominal level to which basically everything audio is related to.
Headroom: The space between a nominal signal (in this case, line level) and the point where the circuit fails. In digital, basically anything under full scale (-0dBFS) would be considered headroom. In analog, it's the space between 0dBVU and the point where the circuit clips (failing completely). In analog, there can be a big difference between "headroom" and "USABLE headroom." We'll get into that in a bit.
Steak: From the old Norse "steik" meaning "roast" -- A slice of meat, typically beef, usually cut thick and across the muscle grain and served broiled or fried (thank you, Wikipedia).
So - You have a microphone and a preamp going into a converter or sound card. Those converters are calibrated at LINE LEVEL. In most cases, over the last several years, most I've seen are calibrated to -18dBFS = line level (or 0dBVU). In other words, if you run a steady signal (a sustained note on a keyboard for instance) through a preamp and turn up the preamp gain until the VU meter reads 0dBVU, at the converter (and on the active track in whatever program you're using the record), it will read -18dBFS (or -18dBFS(RMS) -- full scale, but measured over time).
THIS IS WHERE YOUR GEAR IS DESIGNED TO RUN. This is where it's spec'd at. You will have a decent amount of headroom, the lowest distortion, the best signal to noise ratio, etc., etc., etc. around this level or lower. Some gear - usually very high-quality stuff, has a good amount of usable headroom above this level. A lot of "budget friendly" gear does not. So all of this advice is *more* important if you're using "okay" gear at the input. EVEN YOUR DIGITAL CONVERTERS are ANALOG components up to the converter itself. They don't want to be "beat up" all the time either.
Let's look into headroom -- Above that 0dBVU/-18dBFS range, digital headroom is simple -- Perfect, perfect, perfect, perfect, CLIP. The signal is "what it's supposed to be" until the point of failure. Analog gear (your preamp, compressors, outboard signal processors, etc.) isn't like that... It's more like "Perfect, a bit unfocused, a little noisy, "tight" sounding, spectrally distorted, CLIP. The converter's job is simple - Reproduce the signal it's fed digitally - whether that signal is clean and dynamic or distorted and squishy. The analog chain's job is anything but -- Typically, you're adding 20, 30, maybe 50dB of gain to the incoming signal with a mic preamp. The preamp is working - not just "passing" the signal. And that signal can start to suffer from noise, distortion and dynamically dependent (varying along with volume) spectral imbalance (a skewing of the overall spectrum from an EQ standpoint).
In other words, a nice, thick, chunky guitar tone (for example) might have different characteristics depending on how hot the signal is. The highs might be open and airy and then the signal gets loud for whatever reason and the highs either get swallowed up, or perhaps get very harsh and strident. In any case, it's an inconsistency that isn't' there when the levels are more "normal." Even though the analog gear probably has spec'd headroom well above digital's full-scale, it doesn't mean that signal actually has the integrity it should up to that level. Your car isn't happy running *just short* of red-lining RPM's either.
So what happens is simple - A signal is recorded that's too hot (usually to "use all the bits" which again, is a bunch of BS). It overdrives the input chain not unlike a guitar preamp overdriving a Marshall stack (well, not that much, but the premise is the same). Now, after all the other tracks are recorded, ALL of them need to be attenuated by 12, maybe 15dB or more so the mix doesn't clip. Those distorted, spectrally questionable, squishy, noisy tracks all get turned down.
Are you seeing my point yet?
When you take a steak and cook it until it's burnt, it's burnt. If you pour ice cubes all over it, it doesn't make it more rare - It makes it a cold, wet, burnt steak. No matter what you do, it's still burnt. Just like if you record too hot.
But if you cook a steak a little too rare, you can always heat it up a bit later. You can microwave it without it turning into leather. You can pan-fry it for a few minutes and it's still a tasty, savory piece of steak.
Another applicable analogy: You're a photographer and you have a camera with a lens that can take in 51,000 lumens before it'll melt the image sensor. You want to take an intimate photo of a mother and her newborn baby in a small room with rich, deep earth-tone colors. Are you going to light that room with a 50,000 lumen carbon-arc spotlight? Or perhaps a few soft, incandescent lamps and maybe a candle or two might capture the intimacy and emotion a little better...
Point is, when you use up all your headroom right away, you don't get it back by turning it down. It's gone forever. Sure, you can increase mix headroom or the headroom at the buss - But it's not going to make the track less distorted or fix the skewed S/N ratio.
If this all makes sense but you need quick proof, here's your experiment -- You'll need a few Y-cables (let's not get into the technical aspects of splitting a mic signal - It's an experiment) and at least one stereo (2-channel) preamp.
Record a song using as many tracks as you feel fit. The more, the more apparent. You're going to split the mic signal and record each twice simultaneously. KEEP IN MIND that if you're using condenser mics, apply phantom power to only one side. On one channel of the preamp, set the gain so it peaks between -18 and -12dBFS at the converters and record them to odd numbered channels. On the other, set it as high as possible without clipping and record them to even numbered channels. Record some guitars, drums, maybe piano, of course some vocals, keyboards, go nuts.
Set all the odd numbered ("normal" level) channels to unity and toss up a rough mix to a stereo buss - Which should be a piece o' cake. Switch over to the even numbered channels and figure out how much you're going to have to attenuate them all so the main buss isn't clipping constantly. It might be a lot. Could be a 10-15dB cut on all channels before you can even think of starting to do anything else. Send those to a stereo buss. Solo the busses, one at a time, and try to match the levels between the mixes. You'll probably immediately notice that the "normal" mix is much more open, dynamic, airy, clear, clean, with much more "sonic space" between the instruments than the "hot" mix.
Now, add a limiter on the main buss. Run the "hot" mix into it and bring the level up until it starts to obviously distort and fall apart sonically. Then switch over to the "normal" mix - which should now be "rammed" by the same amount. If your experience is pretty much like everyone else's, the "normal" mix is *still* much more open, airy and dynamic with less distortion and more "crankability" than the other.
THE "DUMBED DOWN" VERSION: Stop recording so hot. Instead of trying to get your tracks to peak at -2dBFS, have them peak between -20 and -12dBFS and your recordings will almost undoubtedly sound better. Mixing will be easier. EQ will be more effective. Compression will be smoother, more manageable and predictable. You're in the age of 24-bit digital recording - Relax and enjoy the headroom. Even if your only concern is the volume of the finished product (which would be a shame, but it happens), recordings made with a good amount of headrom are almost undoubtedly better suited to handle the "abuse" of excessive dynamics control. QUIETER recordings have more potential to be LOUD later. It's because they're usually better sounding recordings in the first place.
John Scrip - MASSIVE Mastering - http://www.massivemastering.com
NOTE: This article picked up by Pro Sound Web May.13.2011 --
https://www.prosoundweb.com/ensuring-youre-recording-at-the-proper-levels/
Let me get something out of the way here - I'm going to try to keep this very "fool proof" - I'm not trying to sound or present this very scientifically - This is just the rantings of hundreds and hundreds of posts on a dozen or more audio forums exploding like a volcano recorded with lots of headroom. I just hope to instill a basic understanding of why certain trends and common beliefs are just plain bad. And by the time you're done reading, and perhaps doing a little experimentation based on this, you won't need me to prove it. You'll know it yourself.
Is this a "miracle cure" for bad recordings? Normally, I'd say no. But with the dozens and dozens - Easily now into hundreds of e-mails, phone calls, letters, forum posts and other forms of communication from one after the next about how this advice has completely changed the way they view recording, I figured I should put it down in a centralized location. The sad part, is this should be common sense. To anyone that grew up "on tape" it probably is. To those brought up in 1's and 0's, it might not be so obvious.
So, if you've been struggling with recordings that sound "weak" or "small" or too dense or "just not 'pro' enough" then please, read on. If this is about you, you might think differently soon.
As a mastering engineer, I work on recordings from pretty much every level of experience. A few years ago, I noticed something unusual.
"Ultra rookie" recordings - Those made by people with little or no experience, sounded fine. They didn't know any better, so they didn't have enough rope to hang themselves with. "Pro" recordings sounded fine. They know what they're doing and/or are using gear with obscene amounts of usable headroom (explained later). The "middle of the road" engineers with a year or two - or much, much more experience -- Those are the recordings that sounded "small" and spectrally challenged. So after quizzing these people over months and months, I came up with the following conclusion...
You're probably recording too hot.
And it's absolutely ruining recording after recording after recording. And it's the simplest thing in the universe to correct.
I know, I know - "It says in the manual to record as hot as you can without clipping." Well, I'm going to flat-out call that B.S. and I'm going to back it up with a simple (if not somewhat time-consuming) experiment.
Also as a mastering engineer, let's get something straight -- I don't like the "loudness war" going on - But I'm as guilty as the next in contributing to it. I can't fight it, as much as I try. Hopefully it'll be over some day. HOWEVER - With the quest for LOUD, there are a LOT of engineers out there shooting themselves in the foot before they even know how to aim. They think that tracking loud and mixing loud contributes to a louder recording after the mastering phase. THIS IS ABSOLUTELY UNTRUE and it's generally the best way to make sure that your recording will NOT have the "loudness potential" of the average commercial release.
CLEAN recordings - Recordings made with low distortion and good spectral balance - THOSE are the ones that handle the "abuse" of the mastering phase with flying colors.
This article isn't intended to give you some secret way of making louder recordings. But it will almost undoubtedly give you the ammo needed to make BETTER recordings. And those BETTER and CLEANER recordings are the ones that can be LOUDER recordings in the end.
First, let's get through a little nomenclature -
dBFS: Deci-Bel (one tenth of a Bel) Full Scale -- On the digital recording scale, -0dBFS is the hottest signal you can have. "All ones." Top of the scale, can't get hotter, etc. Always "minus" as you can never go higher - So the reading will always be a specified amount below 0.
Line Level / 0dBVU: Just what it says. Line level. 0dBVU on an analog VU (volume unit) meter. Pro (+4dBu) or consumer (-10dBv) level, it's line level. We can also refer to this as RMS (Root Mean, Squared), or a level over a specific amount of time. You *can* go above or below 0dBVU. It's simply a nominal level to which basically everything audio is related to.
Headroom: The space between a nominal signal (in this case, line level) and the point where the circuit fails. In digital, basically anything under full scale (-0dBFS) would be considered headroom. In analog, it's the space between 0dBVU and the point where the circuit clips (failing completely). In analog, there can be a big difference between "headroom" and "USABLE headroom." We'll get into that in a bit.
Steak: From the old Norse "steik" meaning "roast" -- A slice of meat, typically beef, usually cut thick and across the muscle grain and served broiled or fried (thank you, Wikipedia).
So - You have a microphone and a preamp going into a converter or sound card. Those converters are calibrated at LINE LEVEL. In most cases, over the last several years, most I've seen are calibrated to -18dBFS = line level (or 0dBVU). In other words, if you run a steady signal (a sustained note on a keyboard for instance) through a preamp and turn up the preamp gain until the VU meter reads 0dBVU, at the converter (and on the active track in whatever program you're using the record), it will read -18dBFS (or -18dBFS(RMS) -- full scale, but measured over time).
THIS IS WHERE YOUR GEAR IS DESIGNED TO RUN. This is where it's spec'd at. You will have a decent amount of headroom, the lowest distortion, the best signal to noise ratio, etc., etc., etc. around this level or lower. Some gear - usually very high-quality stuff, has a good amount of usable headroom above this level. A lot of "budget friendly" gear does not. So all of this advice is *more* important if you're using "okay" gear at the input. EVEN YOUR DIGITAL CONVERTERS are ANALOG components up to the converter itself. They don't want to be "beat up" all the time either.
Let's look into headroom -- Above that 0dBVU/-18dBFS range, digital headroom is simple -- Perfect, perfect, perfect, perfect, CLIP. The signal is "what it's supposed to be" until the point of failure. Analog gear (your preamp, compressors, outboard signal processors, etc.) isn't like that... It's more like "Perfect, a bit unfocused, a little noisy, "tight" sounding, spectrally distorted, CLIP. The converter's job is simple - Reproduce the signal it's fed digitally - whether that signal is clean and dynamic or distorted and squishy. The analog chain's job is anything but -- Typically, you're adding 20, 30, maybe 50dB of gain to the incoming signal with a mic preamp. The preamp is working - not just "passing" the signal. And that signal can start to suffer from noise, distortion and dynamically dependent (varying along with volume) spectral imbalance (a skewing of the overall spectrum from an EQ standpoint).
In other words, a nice, thick, chunky guitar tone (for example) might have different characteristics depending on how hot the signal is. The highs might be open and airy and then the signal gets loud for whatever reason and the highs either get swallowed up, or perhaps get very harsh and strident. In any case, it's an inconsistency that isn't' there when the levels are more "normal." Even though the analog gear probably has spec'd headroom well above digital's full-scale, it doesn't mean that signal actually has the integrity it should up to that level. Your car isn't happy running *just short* of red-lining RPM's either.
So what happens is simple - A signal is recorded that's too hot (usually to "use all the bits" which again, is a bunch of BS). It overdrives the input chain not unlike a guitar preamp overdriving a Marshall stack (well, not that much, but the premise is the same). Now, after all the other tracks are recorded, ALL of them need to be attenuated by 12, maybe 15dB or more so the mix doesn't clip. Those distorted, spectrally questionable, squishy, noisy tracks all get turned down.
Are you seeing my point yet?
When you take a steak and cook it until it's burnt, it's burnt. If you pour ice cubes all over it, it doesn't make it more rare - It makes it a cold, wet, burnt steak. No matter what you do, it's still burnt. Just like if you record too hot.
But if you cook a steak a little too rare, you can always heat it up a bit later. You can microwave it without it turning into leather. You can pan-fry it for a few minutes and it's still a tasty, savory piece of steak.
Another applicable analogy: You're a photographer and you have a camera with a lens that can take in 51,000 lumens before it'll melt the image sensor. You want to take an intimate photo of a mother and her newborn baby in a small room with rich, deep earth-tone colors. Are you going to light that room with a 50,000 lumen carbon-arc spotlight? Or perhaps a few soft, incandescent lamps and maybe a candle or two might capture the intimacy and emotion a little better...
Point is, when you use up all your headroom right away, you don't get it back by turning it down. It's gone forever. Sure, you can increase mix headroom or the headroom at the buss - But it's not going to make the track less distorted or fix the skewed S/N ratio.
If this all makes sense but you need quick proof, here's your experiment -- You'll need a few Y-cables (let's not get into the technical aspects of splitting a mic signal - It's an experiment) and at least one stereo (2-channel) preamp.
Record a song using as many tracks as you feel fit. The more, the more apparent. You're going to split the mic signal and record each twice simultaneously. KEEP IN MIND that if you're using condenser mics, apply phantom power to only one side. On one channel of the preamp, set the gain so it peaks between -18 and -12dBFS at the converters and record them to odd numbered channels. On the other, set it as high as possible without clipping and record them to even numbered channels. Record some guitars, drums, maybe piano, of course some vocals, keyboards, go nuts.
Set all the odd numbered ("normal" level) channels to unity and toss up a rough mix to a stereo buss - Which should be a piece o' cake. Switch over to the even numbered channels and figure out how much you're going to have to attenuate them all so the main buss isn't clipping constantly. It might be a lot. Could be a 10-15dB cut on all channels before you can even think of starting to do anything else. Send those to a stereo buss. Solo the busses, one at a time, and try to match the levels between the mixes. You'll probably immediately notice that the "normal" mix is much more open, dynamic, airy, clear, clean, with much more "sonic space" between the instruments than the "hot" mix.
Now, add a limiter on the main buss. Run the "hot" mix into it and bring the level up until it starts to obviously distort and fall apart sonically. Then switch over to the "normal" mix - which should now be "rammed" by the same amount. If your experience is pretty much like everyone else's, the "normal" mix is *still* much more open, airy and dynamic with less distortion and more "crankability" than the other.
THE "DUMBED DOWN" VERSION: Stop recording so hot. Instead of trying to get your tracks to peak at -2dBFS, have them peak between -20 and -12dBFS and your recordings will almost undoubtedly sound better. Mixing will be easier. EQ will be more effective. Compression will be smoother, more manageable and predictable. You're in the age of 24-bit digital recording - Relax and enjoy the headroom. Even if your only concern is the volume of the finished product (which would be a shame, but it happens), recordings made with a good amount of headrom are almost undoubtedly better suited to handle the "abuse" of excessive dynamics control. QUIETER recordings have more potential to be LOUD later. It's because they're usually better sounding recordings in the first place.
John Scrip - MASSIVE Mastering - http://www.massivemastering.com
NOTE: This article picked up by Pro Sound Web May.13.2011 --
https://www.prosoundweb.com/ensuring-youre-recording-at-the-proper-levels/