|
|
Post by popmann on Feb 24, 2014 19:05:30 GMT -6
Digital full scale does NOT equal analog levels. ie 0dbfs does not equal analog 0db(vu). There's a REALLY long thread about this at the Purple site, but after having yet another person complain about how to "deal with" my "crazy low levels"...I just have to vent--I mean teach. I'm going to put the short punch list rather than wax on and confuse:
--find out what level your converter is calibrated for ANALOG 0db (RME -13dbfs...Burl -12dbfs--defaults, both adjustable)
--that's the level "typical peaks**" should hit...that level PLUS 4db... (so, if -13dbfs=0db...typical peaks should hit between -13 and -9dbfs)
--occasional peaks*** can go over...so long as not over 0dbfs (which you likely knew)
--this has NO relationship to how "loud" your mix or master is--meaning even if you want a finished product that's "competitively loud" (DR4-?-Whatever the F that means to you)...this has NO bearing on that.
--this WILL allow you to make better recordings with the converters you HAVE. Expensive converters just have more ANALOG headroom...ie, they let you track hotter without sounding like poo. This is good-cause sometimes performance levels change and flexibility in what it can handle IS good...and ALSO not necessary for good sound.
--if you get tracks from someone who has cut louder...first thing after import? Trim down to where the levels are where they should be.
--there IS slight exception for those mixing to analog desks...it's actually on the playback/DAC level optimization end rather than recording level...but, let's not confuse that situation with the one a majority of reader will be in--mixing digitally.
In order to not wax on--I will stop there.
Definitions:
**typical peaks...means in the course of the song...the strum strum strum...the snare on the 2/4...the louder end of a vocal line...the pulse of the song should bounce up to some where between those levels.
***occasional peaks....this is that one loud vocal note...where the bass player smacks his strings with a snare hit or flubs a note...a loud drum fill....the 5sec climax of the entire piece, etc...
|
|
|
|
Post by mobeach on Feb 24, 2014 19:58:59 GMT -6
|
|
|
|
Post by jazznoise on Feb 24, 2014 20:29:24 GMT -6
I always record at 24 bit with the peaks aimed at -18. Am I alone in this?
I was at 16 bit I'd probably go a little higher, but I always considered the extra 6 bits in 24 as spare headroom.
|
|
|
|
Post by RicFoxx on Feb 24, 2014 20:40:33 GMT -6
High levels are very bad with average converters and I think most musician engineers don't understand this but in fairness a lot of these cats came from either the analog realm or 16 bit digital where high levels are good. Also people that are working with consumer all in one interfaces monitor and mix through headphones and the headphone amps do not push good headphones so they push the levels way up to hear the mix loud enough.
-18 with peaks around -12 for me...sometimes I push the Burl for the sound to -6
|
|
|
|
Post by svart on Feb 25, 2014 8:57:44 GMT -6
Depends on the program material. If it's something percussive, I'll let it peak around -12 to -6 or so. if it's something more steady like keyboards, I'll let it peak up to something higher, but not clipping.
Honestly, I don't fret about it much. I just learned where my converters like to sit, and where they distort and I'll do level tests before every recording and set my levels for a few dB short of clipping level. My converters don't have soft clipping on the input, so they distort very loudly and very brashly. it's not a good sound at all and easily noticed and fixed.
Sometimes clean guitar is the worst, I'll have audible converter distortion even though the peak will only show around -12. I can only assume it's intersample peaking or something, but I'll have to back off the guitar a little bit to not have audible distortion. Drums, surprisingly, aren't that bad at all. I'll run them up as hot as they'll go without converter distortion at the loudest that the player will play and then back it off a little more.
I think it's a little overrated to think about levels in such detail. It's usually only the newbies that sweat it so much, and it's usually because they read in forums how dreadfully important it is to get high enough levels... So they'll run their meter peaks right up to red.. Not realizing that the latency in their metering probably means many many overs are happening.
That and each DAW is different. I've gotten material from other DAWs that I've seen NOT clip natively, yet be imported into another DAW and it clips like mad..
As I exit the converters and go to the desk, I rarely bother with the levels once again, as they are pretty spot-on during tracking, but sometimes I'll do a sum ITB instead of on the desk and I'll have to drop the levels down a bit to keep from clipping the desk. thankfully I have analog clipping LEDs for this and don't have to rely solely on the meters.
|
|
|
|
Post by popmann on Feb 25, 2014 12:35:07 GMT -6
So, I just want to be clear what you're suggesting....is that a musician....like say our host...should "learn where their converters like to sit....edit...on various program material"?
Rather than simply set the conservative levels that the people who designed the line amp recommend and know that at any given point, you're getting 99% of what that can give? Instead, they should rely on their new to recording ears...to simply set the level wherever sounds good to them.
There's nothing to "sweat" here at all. Look up what your converters are intended for...use your built in PEAK HOLD digital ballistics to make sure you're give or take in that range. Do you disagree with that?
|
|
|
|
Post by jazznoise on Feb 25, 2014 12:44:33 GMT -6
I forgot one caveat.
On room mics were I feel noise might be an issue, I'll crank the gain. Less Johnson noise, so by running a little hotter you keep the noise floor down. Since with ambient mics you get a more averaged sound, it's usually pretty safe.
|
|
|
|
Post by svart on Feb 25, 2014 12:45:25 GMT -6
I'm simply suggesting that getting into too much detail tends to make the newbies who read the details try to follow them to the letter, which leads to more issues than not. I've done it myself back in the day, and I see folks doing it all the time on the purple site all the time.
For example:
Lets say on a specific source, my Alphalink converters will peak around -12 on the edge of hard clipping(due to no soft clipping), and I say to the world "you must set your peak levels to -12 for best sound" and yet when they go and set levels on their Lynx (with soft clipping), -12 actually clips slightly on a similar source, but is not obviously noticeable..
Now you've given advice that someone is following to the T, yet is not correct for their situation, but they're newbies, so they probably don't know what the issue sounds like, until much later, and that means they've wasted a lot of time following an internet "rule".
So yes, I suggest to learn where their converters work best in their setup.
|
|
|
|
Post by svart on Feb 25, 2014 13:00:27 GMT -6
I forgot one caveat. On room mics were I feel noise might be an issue, I'll crank the gain. Less Johnson noise, so by running a little hotter you keep the noise floor down. Since with ambient mics you get a more averaged sound, it's usually pretty safe. Dynamic or condensor? Johnson noise is roughly constant, but you'll be swamping that noise with shot or flicker noise from cranking the preamp, and of course NF from the preamp as a whole.. So one, for another I suppose. The good news is that mics generally have pretty clean gain compared to preamps, so their NF will swamp the preamp NF partially and as long as you don't really bugger up the gain staging from that point onwards, you're going to be OK. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 25, 2014 13:00:58 GMT -6
my Alphalink converters will peak around -12 on the edge of hard clipping Something must be going on related to gain staging elsewhere in the chain. I've never heard of a converter that's not perfectly clean right up to the point of hard clipping at 0 dBFS. Even with cheap sound cards the distortion varies little over a range of input and output levels. I tested this last year and proved (to my satisfaction) that the level you record at is more or less irrelevant as long as you're not down in the noise or above 0 dBFS: The Truth About Record Levels--Ethan |
|
|
|
Post by svart on Feb 25, 2014 13:13:53 GMT -6
my Alphalink converters will peak around -12 on the edge of hard clipping Something must be going on related to gain staging elsewhere in the chain. I've never heard of a converter that's not perfectly clean right up to the point of hard clipping at 0 dBFS. Even with cheap sound cards the distortion varies little over a range of input and output levels. I tested this last year and proved (to my satisfaction) that the level you record at is more or less irrelevant as long as you're not down in the noise or above 0 dBFS: The Truth About Record Levels--Ethan That's for specific sources. Most I can run right up to the DAW meter shows a clip, and beyond, before I hear it. I do think those specific sources have a good amount of super fast transients that DAW meters don't catch. DAW metering is really dependent on latency and CPU usage, with most DAWs erring on the side of using most CPU time for capturing the audio streams, rather than driving meters.. So yeah I totally bet that in those circumstances what I'm seeing is peaking between meter refreshes. |
|
|
|
Post by jazznoise on Feb 25, 2014 15:55:43 GMT -6
I forgot one caveat. On room mics were I feel noise might be an issue, I'll crank the gain. Less Johnson noise, so by running a little hotter you keep the noise floor down. Since with ambient mics you get a more averaged sound, it's usually pretty safe. Dynamic or condensor? Johnson noise is roughly constant, but you'll be swamping that noise with shot or flicker noise from cranking the preamp, and of course NF from the preamp as a whole.. So one, for another I suppose. The good news is that mics generally have pretty clean gain compared to preamps, so their NF will swamp the preamp NF partially and as long as you don't really bugger up the gain staging from that point onwards, you're going to be OK. Either. Johnson noise in a preamp will reduce as the resistive load is decreased. It follows that more gain is always quieter until the signal clips, at which point your SNR will begin reducing again. Taking the Douglas Self' example a 200 Ohm load gives you an absolute noise floor of -129 dB. Clean gain of 70dB - digital or otherwise, makes that -59dB. However pre-amps with a large quantity of shunt or series feedback will be adding noise, in the example schematic he gives has a minimum gain of 0 and a noise floor of -99.6 dB. That's 30dB noisier by the time it hits your daw, and while different circuits will do it differently, you can bet the noise floor drops with increasing gain. |
|
|
|
Post by LesC on Feb 26, 2014 14:36:33 GMT -6
my Alphalink converters will peak around -12 on the edge of hard clipping Something must be going on related to gain staging elsewhere in the chain. I've never heard of a converter that's not perfectly clean right up to the point of hard clipping at 0 dBFS. Even with cheap sound cards the distortion varies little over a range of input and output levels. I tested this last year and proved (to my satisfaction) that the level you record at is more or less irrelevant as long as you're not down in the noise or above 0 dBFS: The Truth About Record Levels--Ethan Ethan, thank you for posting the results of your test, it's something I never seem to find the time to do. Are those tracks for "Ain't No Mountain High Enough" still available anywhere? I googled but couldn't find them. |
|
|
|
Post by joseph on Feb 26, 2014 15:03:09 GMT -6
I like -18dbfs average with only occasional peak transients from drums not ever above -10dbfs. Easy to remember and works for everything.
|
|
|
|
Post by svart on Feb 26, 2014 15:07:57 GMT -6
Dynamic or condensor? Johnson noise is roughly constant, but you'll be swamping that noise with shot or flicker noise from cranking the preamp, and of course NF from the preamp as a whole.. So one, for another I suppose. The good news is that mics generally have pretty clean gain compared to preamps, so their NF will swamp the preamp NF partially and as long as you don't really bugger up the gain staging from that point onwards, you're going to be OK. Either. Johnson noise in a preamp will reduce as the resistive load is decreased. It follows that more gain is always quieter until the signal clips, at which point your SNR will begin reducing again. Taking the Douglas Self' example a 200 Ohm load gives you an absolute noise floor of -129 dB. Clean gain of 70dB - digital or otherwise, makes that -59dB. However pre-amps with a large quantity of shunt or series feedback will be adding noise, in the example schematic he gives has a minimum gain of 0 and a noise floor of -99.6 dB. That's 30dB noisier by the time it hits your daw, and while different circuits will do it differently, you can bet the noise floor drops with increasing gain. Interesting. On the other hand, as the value of an input resistor(roughly source impedance) rises, it's johnson noise contribution goes up, however the current noise in the amplifier also rises and eventually overcomes the johnson noise of the resistors and then considerations for low current noise components has to be made. |
|
|
|
Post by jazznoise on Feb 26, 2014 16:10:08 GMT -6
Current noise is indeed the problem you can run into on the other side of impedance. But, while audio impedance can be low, I've never heard of a circuit in which current noise was the dominant noise source. I'd love to see an exception, though. Every day's a school day.
|
|
|
|
Post by cowboycoalminer on Feb 26, 2014 19:46:48 GMT -6
I always record at 24 bit with the peaks aimed at -18. Am I alone in this? I was at 16 bit I'd probably go a little higher, but I always considered the extra 6 bits in 24 as spare headroom. I track a little hotter than that. Use to I didn't but I've bumped it up some for my particular rig. I find that for me, tracking around -9 to 12 on peaks yields my best result. My RMS is prob around -14 or 15. I get a better signal to noise ratio this way and still have plenty of headroom for mixing. If I need to, I can always trim later. For me, tracking hotter has nothing to do with bit depth, rather noise. **Disclaimer** I would not recommend following this aforementioned technique. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 27, 2014 14:53:18 GMT -6
Are those tracks for "Ain't No Mountain High Enough" still available anywhere? I googled but couldn't find them. I got them from a friend. I have no idea where to find them online. If you email me from my Home Page I can send them to you somehow. (I assume that's illegal, but I doubt it will affect anyone's record sales.) --Ethan |
|
|
|
Post by LesC on Feb 27, 2014 16:12:21 GMT -6
Are those tracks for "Ain't No Mountain High Enough" still available anywhere? I googled but couldn't find them. I got them from a friend. I have no idea where to find them online. If you email me from my Home Page I can send them to you somehow. (I assume that's illegal, but I doubt it will affect anyone's record sales.) --Ethan Oops, sorry Ethan. When you said you 'used the 8-track "master" of the old Motown hit Ain't No Mountain High Enough that made the rounds a while ago', I though you meant they were made public. If you think it's illegal, I wouldn't ask you to give them to me. By the way, nice cat! |
|
