dB Levels

Discussion in 'Recording Studio' started by Tweaker, Oct 24, 2015.

  1. Tweaker

    Tweaker Senior Member

    Messages:
    2,422
    Likes Received:
    1,535
    Joined:
    Jan 18, 2012
    I was doing some reading last night about mixing and one article was talking about the different dB levels, and what's "standard" and all that jazz. Something about -18dB and -12dB, I don't remember specifics of what was written.

    So to achieve the "standards," do you set your track volume to -18 or -12dB (whichever) and then set the input gain on the audio interface?

    Or is this stupid and not how you're supposed to do it?
     
  2. Freddy G

    Freddy G V.I.P. Member

    Messages:
    9,512
    Likes Received:
    18,011
    Joined:
    Mar 18, 2007
    there's a whole lot of missing information in your query....:hmm:
     
    1 person likes this.
  3. Tweaker

    Tweaker Senior Member

    Messages:
    2,422
    Likes Received:
    1,535
    Joined:
    Jan 18, 2012
    Care to fill in the gaps? Cuz I sure as heck don't know what they are...if I did I likely would have included those details.

    EDIT: Nevermind, found some help.
     
  4. yeti

    yeti Senior Member

    Messages:
    3,360
    Likes Received:
    3,169
    Joined:
    Aug 19, 2008
    my pet peeve, hahaha.
    I recently took possession of a SoundDevices 664 mixer/ recorder (a fantastic piece of equipment btw) and I was surprised to see them changing reference levels on us w/o notice. In the words of Tom Waits "the large print giveth and the fine print (specs) taketh away"
    Anyway, there was a time when every piece of gear was analog and in those days there were multiple reference levels, -10 dBV consumer(.3Volts, usually RCA connectors) and +4dBu (1,23Volts, usually XLR connectors) being the most important ones. Think of it this way, the reference level is the level when a VU meter reads "0 VU". Above 0 VU you had the so-called "headroom" the amount of level above reference that could be handled by the circuitry without exceeding a set level of THD (total Harmonic Distortion". You'd be
    surprised at how much headroom some of the old analog gear had, our old Soundcraft console had 33dB of headroom at the input (+37dBu max input)
    Now enter digital
    Truth be told there is no headroom in digital audio (there can be almost infinite headroom within the box but very little at the I/O), there is only 0dBFS (dBFullscale) and below, so the industry decided on a few arbitrary digital reference levels to equal analog 0 VU (-10dBv or +4dBu). Those ref levels are expressed in negative dB numbers such a -20, -18, -14, -12, etc.) and they specify the digital headroom at the I/O, assuming you have I/O capable of "keeping up"
    Unless you work in broadcast there is no "right" digital reference level but the closer the negative number gets to zero the less "headroom" your system has. -20 dBfs seems to me a minimum to use when recording anything.
    However there's one more thing.
    These days the analog I/O on many devices can't handle 20dB of analog headroom (that would be +24dBu) at the input so they just decide that analog reference is 0dBu instead of 4dBu or whatever they want, most interfaces with XLR connectors can barely provide 10dB of headroom over 4dBu=0 VU.

    What does all of this mean?
    Record at levels that allow for levelchanges of +20dB and record at 24bit.
    Why?
    Because 24 bit gets you the extended downward dynamic range to deal with low levels and because you don't want the analog I/O and AD to clip (it might do so at levels as low as 3 or 4 dB above the "pro" analog reference level you might expect when plugging in an XLR connector) so hit the analog input at a very conservative level. Yes, it'll sound anemic unless you have a good piece of gear but that's the price we pay for recording on the cheap.
    In practical applications that might translate into peaks no higher than-12Bb but be careful since interface and ITB digital meters can be all over the place.
    I'd try to get the specs of your Interface and figure out what the max analog input and output level is, then deduct 20dB from that number and then see what the digital reference in the DAW is set at and do the math as to what the interface/DAW turns a -20dB below analog clipping signal into. For example in a decent homerecording rig the interface might crap out at +14dBu analog input and the DAW is turning an arbitrary 0dBu into -14dBFS, so in order to get 20dB headroom you need to shoot for averages that read -6 dB on an analog meter pre-input (-20dBfs on a digital meter) and peaks of <+14dBu = <0dBfs). However if you use a cheap interface it might crap out at +6dBu , despite having XLR inputs and the DAW will still turn 0dBu into -14dBfs. Now the analog clipping will occur at a signal = -8dBfs on your digital meter despite the fact that you still have 8dB before you hit 0dBfs. So a safe peak level would be -10 or 12 dBfs, hence the recommendation to stay at least 10 -12 dB below 0 dB FS for peaks.


    I hope this convoluted explanation makes sense.

    To recap: In a real pro system the math works like this. Analog reference +4dBu = 0 VU = -20 dBfs
    analog maximum input = 24dBu = 0 dBfs
    safe levels = anything below 0dBfs

    in a prosumer setup you need to accommodate the shortcomings of the analog I/O and the DAW's digital reference to arrive at a safe amount of headroom.
     
  5. Freddy G

    Freddy G V.I.P. Member

    Messages:
    9,512
    Likes Received:
    18,011
    Joined:
    Mar 18, 2007
    Thank you for that reply yeti. Well put!
     
  6. yeti

    yeti Senior Member

    Messages:
    3,360
    Likes Received:
    3,169
    Joined:
    Aug 19, 2008
    Thanks, Freddy
    I'm not sure that I explain everything correctly. I don't know if the DAW turns a given analog level into a given digital level or if the interface does that. DAWs can have variable references (-12 or -20 or whatever) so I assume the DAW exerts control but at the same time the interface does the AD so it seems reasonable to assume that the interface operates at a fixed reference (for example -10 dBV = -12dBfs while sporting pro-looking XLr/ TRS combo connectors or something) and the DAW recalculates it, I dunno.
    My guess is that many homerecording interfaces use a really low analog level as reference and then turn it into whatever the DAW tells them via the digital reference level setting. How much headroom that translates to is anybody's guess.

    What I do know is that in many cases you don't have a clue what your meter tells you. Digital devices with analog style metering are common. I personally would not rely on the interfaces meter to tell me what's going on. Here is an example. Our guys use triCasters that look like they can handle pro levels but a quick look at the manual reveals that the max input is around +2dBu (WTF). How do you record safe levels to some POS like that? You treat it as a consumer level box (-10dBV) and use a hard limiter at around 0dBu (find out what that translates to on any given meter) giving you about 10 dB of headroom, not totally horrible if you have decent dynamics processing available in your mixer output.
    The key here is that the recording chain only has as much headroom as the weakest link and usually that's the I/O. Match your levels to it, not some generic rule from the internet because if you have real pro gear there's no reason to not record using the full digital scale. Contrary to popular belief there is a benefit to running levels at a robust, if not hot level especially with cheap I/O (that's where good external dynamic processors come into play) Undercranked safe cheap I/O sounds thin and noisy IME and having a theoretical 145 dB of signal to noise due to 24 bit isn't going to change the fact that your source sounds bad because you didn't gain up enough. Who knows what the linear operating range of a mic/ line input found on a $200.00 interface really is, my guess it's only a small window and it's probably within a few dB below clipping. IME cheap I/O sound worse when you use them conservatively, they seem to excel at recording quasi-static levels close to clipping, rendering them useless for real world material, but maybe I'm wrong here. In the end you have to run safe levels, that's why I called it "the price we pay to record on the cheap"
    But you might be surprised at what some cheap I/O can do.
    For example the Zoom H6 recorder can handle a relatively impressive +22dBu at it's input so at a reference of -20dBfs = 0VU with an external preamp capable of delivering a clean output >22dBu I can safely record peaks of -2dBfs, not -12 or -18. The external preamp will sound better if properly driven, the analog I/O will sound better if properly driven. Conversely if you have a max input of +14dBu at the I/O and drive that external preamp properly I'd keep that setting the same but use a 10dB pad at the I/O instead of undercranking the micpre.
    Note that we haven't even discussed using internal micpres yet. That's a whole other can of worms when looking at levels. There's a reason that there are no useable markings on those knobs, they'd have to lie because if you approach it like I do and say "This mic on speech needs around 50dB of gain to make level" you'd find that the internal micpre can't deliver 50dB of clean gain and that if it could it's output and subsequent analog circuitry couldn't handle the resulting levels. So they keep it murky with unlabeled knobs and mystery meters.
    And the community reacts by stating, don't record at hot levels, no need to (24bit to the rescue) keep it safe at peaks of -12 or whatever. It is a good general approach but when I have good gear to work with I record at levels the analog gear is comfortable at, knowing that my digital chain can deal with it.
     
  7. Tweaker

    Tweaker Senior Member

    Messages:
    2,422
    Likes Received:
    1,535
    Joined:
    Jan 18, 2012
    Very informative responses...gives me a lot more reading to do. Thanks a ton yeti!
     
  8. blues_n_cues

    blues_n_cues Senior Member

    Messages:
    1,802
    Likes Received:
    1,005
    Joined:
    Jun 29, 2010
    -9 to -6db per track & -3db on the Master track sounds & works fine on my system.
    if I want to shop it out for Mastering then I'll drop the Master Track Output to -12 & let that guy deal w/ the loudness wars. :laugh2:
     
  9. LPSGME

    LPSGME Senior Member

    Messages:
    3,747
    Likes Received:
    2,994
    Joined:
    Oct 1, 2008
    I just go by that song... 99 LUFS EBUs.
    bada boom! :naughty:
     
    2 people like this.
  10. yeti

    yeti Senior Member

    Messages:
    3,360
    Likes Received:
    3,169
    Joined:
    Aug 19, 2008




    I think I need to separate the 2 issues in the OP, reference level and recording level.
    A reference level is a standard that ensures interconnectivity between gear, interoperability, etc. Think of it as a standard, kinda like the dimensional lumber standard that a “2X4” (0VU) is actually 3/4" x 1 ½ in size (-10dBv/0dBu/+4dBu).A 2X4 comes in one reference size while 0VU comes in various “sizes” but they are standardized. Those analog reference levels are equivalent to voltages (.3V/.775V/1.23V).If everything is calibrated then you can send a 0VU signal from your mixer to your recorder and at unity gain have a 0VU readout at the recorder as well. This is true for a mixer operating at -10dBv as well a +4dBu, assuming that the receiving device is also a -10 or +4 device. If you plug a -10 signal into a +4 device then the VU on the +4 device will read -11 (it’s not quite -14 because of the way dBu and dBv are measured) and if you plug a +4 mixer into a -10 recorder then the VU at the recorder will be pinned to the red and distortion will occur. So you can see the need for a reference that equipment is rated at. There has always been other levels as well, for example the headroom of a classic compressor like the UREI 1176 was given as 20dB as measured against .775Volts (0dBu) but in general the world lived by -10 (consumer) and +4 (professional) Recently we’ve seen pro-level devices that are using 0dBu = 0VU and that has made things a bit confusing but not too much.

    Before I go on to digital reference I’d like to talk about “converters” of a different kind, called tape machines.
    In case of an analog tape machine the record head converts AC signals into magnetic flux. So far so good but how much flux should equal 0VU? That’s were references come in. A recorder using Quantegy456 tape might be calibrated to turn 0VU into 250 nWb/m. 456tape is called +6 tape because it should be lined up at +6 dB over 185 nWb/m so you can see the same kind of thinking, a reference expressed as a positive or negative number in dB relating to a fixed point
    (in electric signals 0dBu or 0dBv and with flux 185 nWb/m). If you abide by proper calibration then a +6 tape calibrated to 250nWb/m will play back a 0VU tone at 0VU on any tape machine calibrated the same way, you have achieved compatibility and that’s the goal.
    So now there is one more thing and that’s the level that tape is rated at for maximum flux level. In the case of 456 that value is 370 nWb/m, above which “undesirable” distortion will set in. So the range between 250nWb/m and 370 nWb/m is your “magnetic headroom”. Need more headroom? Calibrate to a level below 250 for 0VU, but that comes at a price, more noise. Want less noise? Calibrate at a value above 250 but the price you pay is less headroom.

    Now enter digital reference.
    It’s really simple. We don’t deal with different media that has limitations like tape where the max level of flux is based on how the tape is formulated. In a sense we only have one formula of “tape” and we see that 456’s 370 nWb/m maximum is the equivalent of 0dBFS. Only difference, going above 370nWb/m at the “AC/Flux converter”/ record head might yield pleasing artifacts, going over 0dBFS at the AD will yield awful digital clipping. The digital reference expressed in negative numbers like -12, -14,-18,-20 to mention a few is the equivalent of deciding at what level to line up your tape machine head. Remember the goal was to set a level of flux to correspond to 0VU (whatever that happens to be is decided by the analog ref) same with digital, just pick a negative number that will represent 0VU and define your digital headroom. The closer to zero you go, the less headroom in the digital domain you have at the I/O.
    Now note that none of this has anything to do with recommended recording levels.
    Back to analog tape: We have set our machine to allow for a certain level of magnetic headroom above a level that represents 0VU. The question is now whether the electronics on the machines I/O (beginning to sound familiar?) can handle those levels. Yes, you can slam that needle into the red on a guitartrack recoded on a MTR and saturate the snot out of it but can the mixer deliver those levels to begin with? How about that compressor you have in line, can it handle those levels on the input? Can the tape machine’s front end handle it? How about the playback head and output electronics?
    Those are age-old considerations and nothing new due to digital, you just deal with a different conversion process.
    Now talking about recording levels, at some point the conventional wisdom was that you should not waste bits and record as hot as possible. That school of thought still hasn’t arrived at the trash heap where it belongs.
    In 24 bit recording we can afford to throw away 8 bits and still have CD dynamic range. No problem there. It’s in that context that those who are experts recommend against this approach and that’s where recommended peaks of -12 or -18 or whatever come from. This has nothing to do with reference.
    The 2 arguments for recording at conservative levels are as follows.
    Recording hot causes clipped files and the levels acquired that way are too hot for plug in emulations of classic gear.
    I agree with #1 because as we have seen the entire chain has to be able to deliver a clean signal at those hot levels and many times that is not the case and the specs are unknown.
    The second reason escapes me a bit because if we look at the aforementioned 1176 compressor (arguably one of the most emulated compressors) it can handle levels of 20dB above its analog reference of 0dBu (.775V) so why an emulation couldn’t handle a signal 20 dB above digital reference (that would be 0dBFS in a -20 digital system) is beyond me. I think what that argument really refers to is mixing/ bussing levels. Due to clever math and processing you can have digital “overs” without distortion once inside the box. The SSL C10 audio console has a mindboggling 48,000 dB of internal headroom, so you can jack up that digital gain, tweak that EQ (adds gain too) and run every channel fader, group fader and sub master at full (maybe +12) and still deliver a pristine clean signal at the output. BUT if that where a DAW those levels would be insane for use with a plug in emulating the 1176. You could simulate, I dunno, levels of +100dBu and that would indeed melt the components of a hardware specimen 1176.
    But capture at levels up to 0dBFS are not the problem here.

    I am looking at recording levels based in the hardware I’m using. If I have a MOTU 8pre I know I can’t set the same gain on a mic as I can on a Euphonix S5 to tape via AES. Know your path and then decide. Also keep in mind that different sources demand vastly different headroom, a hihat doesn’t typically even move a VU meter while the transients are in excess of 30dB above the RMS reading. Needless to say you’d record that with very safe levels.
    Lastly you need to know what the postproduction needs are. I frequently record for projects that never see audio finishing so I better record levels close to where they need to sit in the mix otherwise the editors will do stupid things to it.

    I hope this helps.
     
    1 person likes this.
  11. Freddy G

    Freddy G V.I.P. Member

    Messages:
    9,512
    Likes Received:
    18,011
    Joined:
    Mar 18, 2007
    A gold mine of info from yeti here.

    This thread (or just the pertinent info) should be a sticky!
     
    1 person likes this.
  12. yeti

    yeti Senior Member

    Messages:
    3,360
    Likes Received:
    3,169
    Joined:
    Aug 19, 2008
    Thanks, Freddy
    I was just using the opportunity to organize my thoughts on the issue. I hope that I didn't spread falsehoods so if you see any errors in my thinking I'd welcome the correction.
    Regarding recording levels... so which is it?
    My issue with the recommendation of keeping peaks at -12 or -18 is that IME it doesn't sound as good. A great interface allows you to record hotter levels, yet it would also sound good when undercranked but a cheap interface won't allow for hotter levels yet it would benefit from it due to noisy crappy frontend electronics. I wonder if you have any thoughts on this.
    If one would use a cheap interface to record a human voice reading text and the dynamic range was contained at say peaks 6dB above RMS I'd be curious to know if the track recorded at peaks of -12dBFS and 9dB of digital gain added would sound as good as the track recorded with peaks of -3dBFS and no digital gain added. Based on my experience I'd say it wouldn't but I haven't really studied it or done this test.
    Thoughts?
     
  13. Freddy G

    Freddy G V.I.P. Member

    Messages:
    9,512
    Likes Received:
    18,011
    Joined:
    Mar 18, 2007
    That would be a worthy experiment. For myself, I don't have much experience recording with cheap interfaces. At the studio and even in my home set-up it's good AD/DA and pro mic pres and compressors that have boatloads of headroom.

    So the other day when I did that mobile pub recording I spent a bit of time before hand to do some test recordings with the M-Audio Ultra 8. Yes it was a compromise. As I approached the upper limit it got pretty ugly, but if I recorded too conservatively it was as you say "anemic". Since this interface has built in mic pres I don't know what to attribute to the converters and what to the pres. In any case, even at the levels I found to sound the best, it was still kind of pedestrian sounding.

    With the pub band recording I did not record as hot as I normally would....my peaks were somewhere between -12 and -6 depending on the source.
     
    2 people like this.
  14. Tweaker

    Tweaker Senior Member

    Messages:
    2,422
    Likes Received:
    1,535
    Joined:
    Jan 18, 2012
    Definitely the pertinent info!! I can't take any credit for the thread, aside from just being misled by the internet.
     
  15. Wraptail

    Wraptail MLP Vendor

    Messages:
    2,170
    Likes Received:
    1,424
    Joined:
    Jul 23, 2007
    One of the more informative threads I've seen on here in awhile. Thank you for taking the time to share all those insights and experience.
     
  16. John Scrip

    John Scrip Senior Member

    Messages:
    520
    Likes Received:
    575
    Joined:
    Jan 15, 2013
  17. DavidRamey

    DavidRamey Senior Member

    Messages:
    1,466
    Likes Received:
    1,416
    Joined:
    Dec 31, 2013
    I wish this info could be available in a downloadable PDF.
     
  18. yeti

    yeti Senior Member

    Messages:
    3,360
    Likes Received:
    3,169
    Joined:
    Aug 19, 2008
    In the words of Kevin Nealon's weekend update:
    "If you want a transcript of this show, grab a pen and write down everything I say" :laugh2:
     
  19. Playyouraxeon11

    Playyouraxeon11 Banned

    Messages:
    105
    Likes Received:
    42
    Joined:
    Mar 4, 2015
    I think I have an answer, works for me. I use what I have and I learn it by using it. With some stuff, the lights and meters are rubbish, some are not but you try to get all the juice you can. The only way I know how to do it is to do it, same as learning a room, what one corner does, what happens if you drape a few towels around. Trying to assemble one's set up with all that one has to think about is going to be hard so I keep things simple and I keep them cheap. Both Yeti and John Scrip have forgotten more than I know because I don't actually know anything except to plug it in and start. If it sounds sh*te, then it is (same as road testing stomp boxes, most of which are no good for me, too much loss of a good core sound. Really, the only time I liked distortion pedals tended to be when the batteries were giving up).
    I remember (now that I am reminded) about analogue being easier in some ways because it's got better headroom and will tolerate bigger dynamic swings. When going for digital results, it can sound a bit "compressed", all happens the same level or the "presence" fades if you back it off too much. I've never used digital gear that is good enough to overcome this though I'm sure good stuff does.
    So that's the other thing I learned - I can't spend thousands on gaining a bit more headroom and some of my recordings will suffer (which, when it's digital, is a car crash, unlike analogue). Sometimes I'll be lucky and I think it is (as John Scrip, I think, said) a crapshoot. If and when the day comes when I have a serious need to make a top-flight recording, I'll have the resources to do it - that's what that need is, and at that time I'll probably offer Yeti and John Scrip a job to do what I don't know how to do and that my ears are sh*te anyway. I'll ask them to engineer and mix whatever mess it was while I, as usual, go for a smoke outside. I can't do this stuff any more than I can drive racing cars - I don't have the talent.
    I use what I have the best I can and I learn the stuff as I go - the "specifications" don't matter, the result does. Nobody likes or dislike something because of the recording. I mean, I heard "Heatwave" by Martha and the Vandellas on the radio the other day and it's about as rough as it comes and, worst of all, the "mix" (done by placing stuff and people in a room, not twiddling knobs afterwards) is way, way, way "off" but that record was a big hit and it's still exciting today. Another old one was the Walker Brothers, a Brit thing from the 60s. Shocking recording quality and a rough backing orchestra but it worked, there's loads of emotion. They buried it in loads of Phil Spector echo. Sometimes it's just the best thing to do - time is tight, you keep screwing up a bit 'cos you can't play it, what to do? Get the drummer to make a big noise and bash on a moment of delay repeat and forget about it. That's old pro stuff and it still works, I think. If it's a bit "messed up" somewhere, record a load of a Formula One presentation off the TV, put it through a pitch shifter a couple of times, shifting it different ways, mix up these three tracks and lay it in where you need it. Then when you do your ending, bring it back and make it the last thing any one hears of your track and if you are David Bowie, it's "art" and if you're George Martin, you're a genius.

    What Yeti posted is incredible. Knowledge, clear as a bell. I can really only suggest "Seat of the pants" and "take no notice" and it doesn't give the picture so well but when it comes to the stuff you have, the most important thing has to be the result when you hear it and it takes practice, for me there's no escaping that. Maybe others can do it from spec sheets.

    I have a table, a Zoom r24, two Tascam VLs5 monitors, a Shure Unidyne, A Rode 3 condenser mic, a Yamaha THR 10, a Roland Microcube, an AC30, a Dunlop Cry Baby, an SG standard, an es335, a Strat, a J45 and a Precision bass. I also know one or two people, which is handy. I absolutely cannot sing, no matter how I try to trick it.

    I also have a Trio band creator. Running a recorded track of any kind into the programming input on this really f*cks it up. Then I can play it back and jam along to it. That's the sort of thing to have on hand if you need to patch or bury something because, even if it's dreadful, there will be a 5% piece of usable "how on Earth did they do that?" on it. I also like to end a song at different times on the different tracks and run this into the intro of the next piece. It means that, periodically, everything goes mad and I like the drums to be played real time, no metronome. Sometimes I start a track with a bit of rhythm guitar and place everything else on afterwards. It drifts all over the place but it doesn't matter to me. I don't say that you should do it but there are no rules.

    My neighbours hate me, so will Youtube. I intend to use some of this material in a bit of advertising. Fun, I'd say. There are no rules.
     

    Attached Files:

Share This Page