Thread: Hopefully not an endless thread...

Results 1 to 20 of 35

  1. #1
    Join Date Nov 2007
    Location Portland, Oregon
    Posts 144
    Rep Power 268435602

    Default Hopefully not an endless thread...

    Hey Guys,
    I've just spent the last 3 hours reading all around the internet about digital levels/metering/gain staging... I've also read some stuff here in the womb.

    Bottom line seems to be that a safe and (some say) better approach
    to ITB tracking is to let your peaks hit, let's say for instance, -18db.

    Ok, well here's the deal. I've recorded some stuff at very moderate levels. But my waveform literally looks like an emergency room tragedy. The patient has flat-lined. A quick look at the peak meter is showing me that this particular track peaks at about -18db.

    I opened the MixIt 3 project I had and looked at those tracks.
    The waveforms aren't huge and wavy, but they have definition to them, you know? Like, you can see each drum hit, etc. With the faders at 0db, they seem to peak around -6db.

    So, before I get chewed out, I realized that I should be mixing with my ears and NOT with my eyes...but I can't help but wonder why my waveform is basically a line, and the mix-it 3 waveforms are not.

    Okay, here's the second question. Suppose you are tracking with peaks hitting somewhere between -18db and -16db. Where the heck is the fader supposed to be at? Should that be at 0dbFS? Or should the fader be down at -18db too?

    Also, something I noticed was that when I pull the raw tracks from the Mix-It 3 into Cubase with the faders at their default of 0db, I get some red-light clipping. I took the master fader down to -4db, and every other track down to -6db and that finally stopped the clipping. What's up with that?

    Any and all help is appreciated!
    I've tried to read up before posting, but if there is something similar already here at the womb, a link to that post would be helpful too!

    Thanks!
    -Daniel
    Last edited by RTLdan; September 13th, 2009 at 12:06 PM. Reason: typos
    " In the beginning was the Sound, and the Sound was with Volume, and the Sound was LOUD.

    And the Sound became Recorded and played among us, and we beheld Its glory, the glory as of the only recorded of the SM57, full of midrange and treble."

    -Quote in a post from Brendo
  2. #2
    Join Date Nov 2006
    Location Sweden
    Posts 14,275
    Rep Power 536871515

    Default Re: Hopefully not an endless thread...

    let your peaks hit, let's say for instance, -18db.
    Actually, on an analog console you probably wouldn't let your peaks hit at 0. They would tend to go a bit over that. You might see levels hovering around 0 dB on your VU meters, but remember analog VU's won't catch the absolute peak.

    However, it pays to know that a popular reference level for 0dBVU on an analog console at the +4 operating level is equal to -18dBFS on a digital machine.

    But my waveform literally looks like an emergency room tragedy.
    Doesn't really matter. I used to worry about looks with my stage wardrobe, not with waveforms.

    I opened the MixIt 3 project I had and looked at those tracks....
    With the faders at 0db, they seem to peak around -6db.
    That was Mixerman's track, right? I know Mixerman likes to track pretty hot to tape, so with properly calibrated reference levels, you would see peaks quite a bit above -18 dBFS (it corresponding to 0dBVU on his console). Also remember that he likely transferred to digital through Radar converters, which have really beefy power supplies and will handle high voltage signals better than your average converter.

    Suppose you are tracking with peaks hitting somewhere between -18db and -16db. Where the heck is the fader supposed to be at? Should that be at 0dbFS?
    You are confusing some measurement units here. 0dBFS is a full-bore 0 dB digital signal. The fader values are simply relative values, where 0 dB is unity gain. The signal is what it is. If the meter is post-fader, your fader should be set to 0dB to read the true signal level. But you should also remember that few meters tell the absolute truth. Bob Olhsson knows way more about that part than I do, and I believe he has posted some stuff about it.

    Also, something I noticed was that when I pull the raw tracks from the Mix-It 3 into Cubase with the faders at their default of 0db, I get some red-light clipping. I took the master fader down to -4db, and every other track down to -6db and that finally stopped the clipping. What's up with that?
    Where did you see the clipping? On separate tracks or on the master fader?

    If you see clipping on individual tracks, it means the original signal is clipped to begin with (and I find that hard to believe, since you noted yourself that they tended to peak around -6 dB).

    However, your Cubase "mixer" is a digital summing device, which means it sums all the tracks to the master output like any mixer. When you sum tracks, the signal energy adds up, so the more tracks you have, the lower each individual track has to be in order to not cause clipping on the master fader.

    By attenuating the master fader, you won't see the clipping, but if your 2-bus processing is pre-fader, it will still be hit with the same signal level, causing the plugins to behave differently.

    If all tracks were recorded at more moderate levels, perhaps even close to the desired relative levels of your mix, you could have all faders at or near unity without any risk of overload.

    As an aside, I was always on the fence about this technique (often referred to as the "broom handle" technique) with analog recording, because it was more about the individual level of the tracks and how they were hitting tape, but with digital, that consideration pretty much goes away as long as you watch your strongest levels in relation to the 2-bus.

    Again, guys like Bob, Mixerman and Weedy will probably have much more information about this.


    otek
    Last edited by otek; September 13th, 2009 at 01:00 PM.
    "Tube color is not the 'thing'. Why would the most linear amplifying device have a color?" - Jonte Knif
  3. #3
    Join Date Nov 2006
    Location Songwriter Gulch
    Posts 10,140
    Rep Power 536871378

    Default Re: Hopefully not an endless thread...

    It's your RMS level that should be around -18. Peak would be around -6.

    In 24 bit digital it's lots better sounding to error on the low side than the high side.
    Bob's room 615 562-4346
    Georgetown Masters 615 254-3233
    Interview
    Artists are the gatekeepers of truth!- Paul Robeson
  4. #4
    Has Many Personal Intergritys Expensive Boat Anchor
    Join Date Nov 2006
    Location New York
    Posts 10,663
    Rep Power 536871395

    Default Re: Hopefully not an endless thread...

    until you hit REALLY REALLY REALLY low level, like nearly none, there is no DOWN SIDE to recording into the DAW at lower levels.

    so take it fwiw that even if you think it only MIGHT be better to leave more headroom, err on the 'too low' side.

    you'll be better for it later.
  5. #5
    Garden variety weasel...has ripped flesh. Prom Queen
    Join Date Jan 2007
    Posts 511
    Rep Power 536871054

    Default Re: Hopefully not an endless thread...

    Ok, well here's the deal. I've recorded some stuff at very moderate levels. But my waveform literally looks like an emergency room tragedy. The patient has flat-lined. A quick look at the peak meter is showing me that this particular track peaks at about -18db.
    Many apps nowadays (Cubase, Logic...) contain a setting or button that allows a "compressed" display of the waveform, probably for this very reason. Please note that this only affects the display, not the sound.
  6. #6
    A Top 10 Beaver Licker Grumpy young man
    Join Date Nov 2006
    Location Nouvelle Zealande
    Posts 1,941
    Rep Power 2500000000

    Default Re: Hopefully not an endless thread...

    so take it fwiw that even if you think it only MIGHT be better to leave more headroom, err on the 'too low' side.

    you'll be better for it later.
    Quoted for emphasis.

    When that vocalist belts out that high note on the only take worth saving, and you peak at -5 you'll be a smiling, happy engineer. If you've bricked it, you'll be making the disappointed, furrowed brow face for quite some time.

    Go quiet. It's 24 bit.
  7. #7
    Join Date Nov 2006
    Location Hobart, Tasmania
    Posts 1,260
    Rep Power 536871082

    Default Re: Hopefully not an endless thread...

    My last long-term audio job involved tracking five VERY dynamic voice actors for an animated TV series. My equipment basically comprised 5 LDCs running straight into the mic pres of a SADiE H64. No dynamics control ahead of the ADC, and even the gain controls were fiddly little on-screen rotary controllers.

    Suffice it to say, riding the gain for 5 performances at once was not going to be a happening thing. I often had to run at much lower average levels than -18, simply to accommodate the extreme peaks that some of the actors generated when they hit the more expressive parts.

    Sound-wise, this was absolutely not a problem. It made editing a little more cumbersome due to the aforementioned flat-line waveform issue. But not anywhere near as annoying as trying to ride 5 gain controls using a mouse would have been
  8. #8
    Owns and wears an Elvis jumpsuit Abstruse Absquatulater
    Join Date Apr 2008
    Location JHB, RSA
    Posts 2,385
    Rep Power 536871101

    Default Re: Hopefully not an endless thread...

    Many apps nowadays (Cubase, Logic...) contain a setting or button that allows a "compressed" display of the waveform, probably for this very reason. Please note that this only affects the display, not the sound.
    I was just gonna say....

    This issue, IMO, is the core idea of audio engineering/mixing. It all starts during tracking and levels you achieve then and progesses into how you structure your gain within your busses and inserts. In the past, I used to just make sure that the levels were there, weren't overloading, and then I'd get a balance with faders later. What I came to realise that it saves a lot of headache, not to mention sounds better, when you get your balance right (especially with the drums) at the input. IOW, when all faders are at zero, you have a relative balance and a fair representation of the mix. Also remember that for every 3dB you attenuate, you lose 1-bit of resolution so keeping your faders at 0 and your gain staging fair (auditioning processing at equal perceived volume levels) goes a long way in retaining the original fidelity of the recording.

    I try to think about every buss and insert point on the mixer as a gate that the audio passes through where the gain might be affected. Focussing on the insert points, if your peaks coming into a particular channel are hitting say, -12, it's best to make sure that that level is maintained on the input and the output of each plugin in succession. IOW, if you bypass all plugins at once, there shouldn't be any volume change. This directly relates to and depends on having your relative levels balanced at the input so that you're not excessively boosting or attenuating gain and hence losing bits.

    To illustrate an example for things like compressors, always bring the threshold down first, even if it looks really low, and keep your eye on the reduction and output meters. Take note on how much reduction is taking place and then use your ears to give a little make up to match the input level. The reason I mentioned this is because you must always be mindful of the levels going in and out of each plugin. Do not be fooled into thinking everything is dandy just because your recorded level was good and the channel output meter is not peaking. Whats going on between every insert point and within each plugin is key here.

    Cheers
    Originally Posted by Slipperman
    Deny everything and claim it's all "Haas-Moeller and Graffenfrimitz" MICING techniques and CAN'T be removed from the tracks without damaging the stereo field due to the "Von Stauffenberg Effect".
    Pan Music Productions
    http://gregbester.xp3.biz
  9. #9
    Join Date Nov 2007
    Location Portland, Oregon
    Posts 144
    Rep Power 268435602

    Default Re: Hopefully not an endless thread...

    Whoa...not even sure where to begin, but I read a lot of good stuff from all of you.

    That makes sense what Bob said about an RMS level of about -18db, not a peak of -18db.

    Otek, you definitely deserve a rep boost for your reply.
    Is it safe to assume that any major DAW will have a preference
    about the 2-buss being pre or post fader? I guess I'll go digging around in Cubase to find out...

    Thanks everyone for your replies!
    I'm going to re-read through the whole thread to absorb the info.

    -Daniel
    " In the beginning was the Sound, and the Sound was with Volume, and the Sound was LOUD.

    And the Sound became Recorded and played among us, and we beheld Its glory, the glory as of the only recorded of the SM57, full of midrange and treble."

    -Quote in a post from Brendo
  10. #10
    Owns and wears an Elvis jumpsuit Owns a kick-start dildo
    Join Date Jun 2008
    Location In a little African country do
    Posts 2,358
    Rep Power 536871099

    Default Re: Hopefully not an endless thread...

    Many apps nowadays (Cubase, Logic...) contain a setting or button that allows a "compressed" display of the waveform, probably for this very reason. Please note that this only affects the display, not the sound.
    Ardour has a configurable logarithmic waveform view, which I use frequently when working. It puts the signal bandwidth in better perspective while recording, and you can see even low signals properly. Pretty sure the other DAWs should have this as well.
  11. #11
    Join Date Nov 2006
    Location Sweden
    Posts 14,275
    Rep Power 536871515

    Default Re: Hopefully not an endless thread...

    remember that for every 3dB you attenuate, you lose 1-bit of resolution
    It's actually every 6 dB.

    With 24-bit, you can peak at -48 dBFS (i.e. your meters would be barely budging) and you'd still have 16 bit resolution.

    Is it safe to assume that any major DAW will have a preference about the 2-buss being pre or post fader?
    I'm not sure about other DAWs, but Logic has its metering post master fader and the inserts pre. This means you could be perfectly in the green on the meters and still be bombing the 2-bus plugins.



    otek
    "Tube color is not the 'thing'. Why would the most linear amplifying device have a color?" - Jonte Knif
  12. #12
    Ducked in here to avoid the paparazzi Child Prodigy Gone Bad
    Join Date Dec 2008
    Posts 5
    Rep Power 106

    Default Re: Hopefully not an endless thread...

    Ardour has a configurable logarithmic waveform view, which I use frequently when working. It puts the signal bandwidth in better perspective while recording, and you can see even low signals properly. Pretty sure the other DAWs should have this as well.
    Yep... The majority of DAWs have Waveform zoom (not to be confused with track zoom)

    Cubase/Nuendo (top right slider I think in the edit window), Logic (bottom right button with the little waveform image), ProTools (I forget where :-() . So that your waveforms can be seen with the definition that you are accustomed to.
  13. #13
    Owns and wears an Elvis jumpsuit Abstruse Absquatulater
    Join Date Apr 2008
    Location JHB, RSA
    Posts 2,385
    Rep Power 536871101

    Default Re: Hopefully not an endless thread...

    It's actually every 6 dB.
    Damn. I couldn't remember if it was 3 or 6dB when I wrote it and evidently chose the wrong value. I won't forget it now. I think the idea is still relevant though in that you don't want to lose bits if you don't want have to.

    Dan, you asked this:

    Is it safe to assume that any major DAW will have a preference about the 2-buss being pre or post fader? I guess I'll go digging around in Cubase to find out...
    If you right click near the meter on any channel on the Cubase window a pop-up menu will appear where you will find a sub-menu for individual channel and global meter settings. Here you can choose what meter configuration you are looking for, i.e. pre/post fader, etc. Also, a good thing to remember is that in Cubase/Nuendo, insert points 1-6 are PRE fader and 7-8 are POST fader.

    Cheers

    - MoDontKnowtheSixFacta
    Originally Posted by Slipperman
    Deny everything and claim it's all "Haas-Moeller and Graffenfrimitz" MICING techniques and CAN'T be removed from the tracks without damaging the stereo field due to the "Von Stauffenberg Effect".
    Pan Music Productions
    http://gregbester.xp3.biz
  14. #14
    Join Date Nov 2006
    Location Sweden
    Posts 14,275
    Rep Power 536871515

    Default Re: Hopefully not an endless thread...

    I think the idea is still relevant though in that you don't want to lose bits if you don't want have to.
    At 24 bit with its (theoretical) 144 dB of dynamic range, it's actually far less critical than the problems associated with hitting your converters too hard.


    otek
    "Tube color is not the 'thing'. Why would the most linear amplifying device have a color?" - Jonte Knif
  15. #15
    Join Date Nov 2007
    Location Portland, Oregon
    Posts 144
    Rep Power 268435602

    Default Re: Hopefully not an endless thread...

    Mo Facta,
    Thanks for the tip for Cubase! + rep for you sir!

    -Daniel
    " In the beginning was the Sound, and the Sound was with Volume, and the Sound was LOUD.

    And the Sound became Recorded and played among us, and we beheld Its glory, the glory as of the only recorded of the SM57, full of midrange and treble."

    -Quote in a post from Brendo
  16. #16
    Join Date Nov 2006
    Posts 5,359
    Rep Power 536871218

    Default Re: Hopefully not an endless thread...

    At 24 bit with its (theoretical) 144 dB of dynamic range, it's actually far less critical than the problems associated with hitting your converters too hard.


    otek
    Actually converters don't ever have all that. Usually you have 20 effective bit resolution, but then 16-bit is never all 16 because of dither, so you'd have something like 90 dB effective range on 16-bit system vs 110-120 dB on 24-bit. Still having -20 dB peaks is better than ideal 16-bit (and that's enough for "brothers in arms" and plenty other records)
    When in doubt, mumble!

    EVERYTHING SOUNDS LIKE SHIT IF YA LISTEN LONG AND HARD ENOUGH.
  17. #17
    Owns and wears an Elvis jumpsuit Owns a kick-start dildo
    Join Date Jun 2008
    Location In a little African country do
    Posts 2,358
    Rep Power 536871099

    Default Re: Hopefully not an endless thread...

    At 24 bit with its (theoretical) 144 dB of dynamic range, it's actually far less critical than the problems associated with hitting your converters too hard.
    +1

    Especially if you think about what happens to your nice 24bit input sources during mixing... If you have 32 tracks, and your target is CD, each can be as low as 11 bits in the mix, while still perfectly balanced in the track.

    Of course each can get 16bit fidelity by itself... but you rarely have a single track at max with nothing else.

    With my classical recordings, I rarely peak higher than -12, and almost never than -6, and classical stuff is hugely dynamic.

    Sounds just fine in the final product.
  18. #18
    Join Date Nov 2006
    Location Sweden
    Posts 14,275
    Rep Power 536871515

    Default Re: Hopefully not an endless thread...

    Actually converters don't ever have all that.
    Hence the word "theoretical". You would also have to take into account the best-case noise floor of even first rate consoles and preamps.


    otek
    "Tube color is not the 'thing'. Why would the most linear amplifying device have a color?" - Jonte Knif
  19. #19
    Join Date Nov 2006
    Posts 5,359
    Rep Power 536871218

    Default Re: Hopefully not an endless thread...

    Hence the word "theoretical". You would also have to take into account the best-case noise floor of even first rate consoles and preamps.


    otek
    And that would make -48dBFS estimate overenthusiastic)))
    I think best bet is having analog feeding digital at 0dBVU = -18dBFS RMS, with peaks no higher than -9 to -12dB.
    When in doubt, mumble!

    EVERYTHING SOUNDS LIKE SHIT IF YA LISTEN LONG AND HARD ENOUGH.
  20. #20
    Join Date Nov 2006
    Location Sweden
    Posts 14,275
    Rep Power 536871515

    Default Re: Hopefully not an endless thread...

    I think best bet is having analog feeding digital at 0dBVU = -18dBFS RMS
    Which is what I said in my first post in this thread.

    Jesus, dude. You're really busting my balls.


    otek
    "Tube color is not the 'thing'. Why would the most linear amplifying device have a color?" - Jonte Knif

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts