The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
But it is impossible to record at -18DBFS RMS Metering & Analysis Plugins
Old 12th January 2017
  #31
I think of it as two simple guidelines:

1) Listen
2) Don't Clip

Listen, becomes there's always something analog in front of the A/D - the mic, the preamp, your voice, the room, whatever. Whenever there's something analog, you should, first and foremost, listen.

Don't Clip, because digital has a hard ceiling.

As to how much headroom you leave, that's up to to you. Digital has so much headroom, it's almost silly to risk recording anywhere close to clipping level as you'll likely not need that much dynamic range. That one perfect take captured with 6 or 10 or 18 dB of headroom is going to sound great and be a keeper. That same perfect performance, digitally clipping just as the performer hits those big payoff passages, might have to be thrown away.

..ant
Old 12th January 2017
  #32
Quote:
Originally Posted by psycho_monkey View Post
Interesting.

No mix engineer I've ever worked with, as I said, listened at one volume. I've even read a quote from Bob Clearmountain saying most of the time, he's listening at lower than speaking volume (I certainly couldn't work all day at levels where you have to raise your voice to get over it), and also that anyone is welcome to change the level at any time. My clients frequently want to change playback levels too!

All that said - I can see why it's a great idea for initial gainstaging and balancing, and I'm curious to try calibrating my room to that setup, at least as an experiment.

It's a long time since I worked at Abbey Road (and I was VERY junior - just a runner, occasional assistant)...I really don't recall anyone talking about this or working like this, and I sat in on some big sessions there (Lord of the Rings, Harry Potter, Gangs of New York recording sessions and the like). Outside of post, you are the first person I've ever heard of to suggest calibrating a room like this. I've never seen a console in a mix studio with a nominal monitoring level marked (and I've worked in many, if not most, of the mid-high level London studios).

For mixing a film - again, I totally get why it's necessary, and for mastering an album too. But when consistent loudness isn't the end goal, I don't think it's necessary or advantageous to listen at the same volume the whole time, even if it is useful for setting up the initial balance.



Quite. But I am curious to try this setup for sure.
When tracking or mixing you will ultimately change the volume as the day goes on. Hitting Dim, dropping the volume even more, maybe even going a little louder than your ref.

But you are frequently returning back to your ref level to make judgment calls. Even though Bob mixes quietly, that doesn't mean he hasn't calibrated the studio to a ref level... it just might be his ref level is 76dB or something like that instead of 85dB.

Technically speaking, 85dB is supposed to be used when you are 11' or farther from the speakers (like soffit mounted speakers for example). I forget who came up with that, It might have been Dolby, or ITU or AES or some organization like that. Our sense of volume gets skewed as the speakers get closer, but nobody has done any studies to see what the idea ref levels are for under 11'. Some use 82dB SPL, some use 79dB SPL, some go down to 76dB SPL or lower.

A lot of studios won't tell you what the calibration is unless you ask. I usually make it a point to ask, when I first start getting settled. Some studios also use -20dBFS = +4dBu while other studios use -18dBFS = +4dBu. So I always ask what everything is calibrated to, just so I know.

If you do calibrate... it takes a couple weeks for your ears and brain to get used to it (try to pick a cal level that is close or exactly the level you usually end up listening at and the adjustment time will be quicker), if you aren't already used to doing it. Listen to a lot music for a few weeks after calibrating. After a few weeks you will have "internalized" what your favorite mixes sound like at the ref level. And then when moving between rooms if you calibrate the room to the same ref, you'll start to notice it's very easy to get sounds and mixes that are consistent with what you do elsewhere. I use -20dBFS = +4dBu = 85dBFS at my studio. so when I go to a studio that is calibrated differently... say -18dBFS = +4dBu = 79dBFS... I know that if I turn the monitors up 6dB I will be hearing what I am used to and I will also know that the system has 2dB less headroom than I am used to so I will be a hair more caution when setting levels for fast transients like snare and tamborine.

Anyway... when you start a tracking session or a mixing session... set the monitor level to the calibrated reference level you chose and have been getting used to... set your tracking levels or get a basic mix going... then you can start changing the monitor levels and using the dim to see what the instruments sound like at different levels.
Old 12th January 2017
  #33
To add just a little. IF you use a VU plug-in, that normally let's you pick a dBFS level for 0 VU. I tend to set -18 dBFS, but while the needle will normally be from -2 to +2, the peak lights (which can correspond to say -8dBFS) will be going off frequently.

I sometimes do live-to-two for broadcast, and this sort of technique is useful there, and I'll either get real physical VUs or send a copy of the digital signal from the mixer to a DAW with the plug-in just for that feature, so that I can do the appropriate gain riding (may raise up the soft passages as well as pull down the loud ones) to remain within the limits of the broadcast.

As others have noted, when tracking - just be sure not to clip anything. When mixing, I try and peak no higher than -3 to -1 dBFS depending on how much headroom I want to leave for mastering.
Old 13th January 2017
  #34
Quote:
Originally Posted by Etch-A-Sketch View Post
When tracking or mixing you will ultimately change the volume as the day goes on. Hitting Dim, dropping the volume even more, maybe even going a little louder than your ref.

But you are frequently returning back to your ref level to make judgment calls. Even though Bob mixes quietly, that doesn't mean he hasn't calibrated the studio to a ref level... it just might be his ref level is 76dB or something like that instead of 85dB.

Technically speaking, 85dB is supposed to be used when you are 11' or farther from the speakers (like soffit mounted speakers for example). I forget who came up with that, It might have been Dolby, or ITU or AES or some organization like that. Our sense of volume gets skewed as the speakers get closer, but nobody has done any studies to see what the idea ref levels are for under 11'. Some use 82dB SPL, some use 79dB SPL, some go down to 76dB SPL or lower.

A lot of studios won't tell you what the calibration is unless you ask. I usually make it a point to ask, when I first start getting settled. Some studios also use -20dBFS = +4dBu while other studios use -18dBFS = +4dBu. So I always ask what everything is calibrated to, just so I know.

If you do calibrate... it takes a couple weeks for your ears and brain to get used to it (try to pick a cal level that is close or exactly the level you usually end up listening at and the adjustment time will be quicker), if you aren't already used to doing it. Listen to a lot music for a few weeks after calibrating. After a few weeks you will have "internalized" what your favorite mixes sound like at the ref level. And then when moving between rooms if you calibrate the room to the same ref, you'll start to notice it's very easy to get sounds and mixes that are consistent with what you do elsewhere. I use -20dBFS = +4dBu = 85dBFS at my studio. so when I go to a studio that is calibrated differently... say -18dBFS = +4dBu = 79dBFS... I know that if I turn the monitors up 6dB I will be hearing what I am used to and I will also know that the system has 2dB less headroom than I am used to so I will be a hair more caution when setting levels for fast transients like snare and tamborine.

Anyway... when you start a tracking session or a mixing session... set the monitor level to the calibrated reference level you chose and have been getting used to... set your tracking levels or get a basic mix going... then you can start changing the monitor levels and using the dim to see what the instruments sound like at different levels.
Certainly makes a lot of sense, and as I said I'm interested to try it.

I've just never seen it done in practice - and I've worked with a fair few "name" mixers, and at 2 of the biggest studio companies in the UK. I was inhouse assistant and I wouldn't have known to answer that, and we had mix rooms belonging to Hugh Padgham and people of that stature.

I'm on a private FB group with a lot of big names...I might ask!

(PS you mean -20dBFS = +4dBu = 85dB SPL, not 85dBFS right? I just noticed that typo when cross posting!)
Old 13th January 2017
  #35
Lives for gear
 

Quote:
Originally Posted by Etch-A-Sketch View Post
Some studios also use -20dBFS = +4dBu while other studios use -18dBFS = +4dBu. So I always ask what everything is calibrated to, just so I know.
Wouldn't you be more interested in the acoustic output than line level electrical output for mixing purposes?
Old 13th January 2017
  #36
Quote:
Originally Posted by Audiop View Post
Wouldn't you be more interested in the acoustic output than line level electrical output for mixing purposes?
Gotta ask about both and understand the whole chain I think.
Old 14th January 2017
  #37
Quote:
Originally Posted by Etch-A-Sketch View Post
When tracking or mixing you will ultimately change the volume as the day goes on. Hitting Dim, dropping the volume even more, maybe even going a little louder than your ref.

But you are frequently returning back to your ref level to make judgment calls. Even though Bob mixes quietly, that doesn't mean he hasn't calibrated the studio to a ref level... it just might be his ref level is 76dB or something like that instead of 85dB.

Technically speaking, 85dB is supposed to be used when you are 11' or farther from the speakers (like soffit mounted speakers for example). I forget who came up with that, It might have been Dolby, or ITU or AES or some organization like that. Our sense of volume gets skewed as the speakers get closer, but nobody has done any studies to see what the idea ref levels are for under 11'. Some use 82dB SPL, some use 79dB SPL, some go down to 76dB SPL or lower.

A lot of studios won't tell you what the calibration is unless you ask. I usually make it a point to ask, when I first start getting settled. Some studios also use -20dBFS = +4dBu while other studios use -18dBFS = +4dBu. So I always ask what everything is calibrated to, just so I know.

If you do calibrate... it takes a couple weeks for your ears and brain to get used to it (try to pick a cal level that is close or exactly the level you usually end up listening at and the adjustment time will be quicker), if you aren't already used to doing it. Listen to a lot music for a few weeks after calibrating. After a few weeks you will have "internalized" what your favorite mixes sound like at the ref level. And then when moving between rooms if you calibrate the room to the same ref, you'll start to notice it's very easy to get sounds and mixes that are consistent with what you do elsewhere. I use -20dBFS = +4dBu = 85dBFS at my studio. so when I go to a studio that is calibrated differently... say -18dBFS = +4dBu = 79dBFS... I know that if I turn the monitors up 6dB I will be hearing what I am used to and I will also know that the system has 2dB less headroom than I am used to so I will be a hair more caution when setting levels for fast transients like snare and tamborine.

Anyway... when you start a tracking session or a mixing session... set the monitor level to the calibrated reference level you chose and have been getting used to... set your tracking levels or get a basic mix going... then you can start changing the monitor levels and using the dim to see what the instruments sound like at different levels.

Quote:
Originally Posted by psycho_monkey View Post
Certainly makes a lot of sense, and as I said I'm interested to try it.

I've just never seen it done in practice - and I've worked with a fair few "name" mixers, and at 2 of the biggest studio companies in the UK. I was inhouse assistant and I wouldn't have known to answer that, and we had mix rooms belonging to Hugh Padgham and people of that stature.

I'm on a private FB group with a lot of big names...I might ask!

(PS you mean -20dBFS = +4dBu = 85dB SPL, not 85dBFS right? I just noticed that typo when cross posting!)
So I had a few replies from some of my fellow engineers on the group I mentioned.

I'm not going to attribute anything to anyone in particular but they're all working engineers in varying fields, some of whom will be familiar, some not.

As I thought, most of the rock guys were kind of "nah - it's a post thing".

One of the classical/film guys (who I actually knew back from Abbey Road during Lord of the Rings) said it's usual anywhere they mix film or TV music. A couple of the more tech-y pop/rock mixers said they have a calibration level, but don't always stick to it - more somewhere just to come back as a reference.

Others didn't do it at all - and I'd be very surprised if in my parts of the world any studio other than the film places (like Abbey Road) did. It's certainly not something I've ever been asked to do or provide info.

One thing I did notice - we had a relatively loud (not painful) playback in here, and I did a quite iPhone measurement, and it was around 82dB SPL. That was far too loud to work at all day for me, and I have a big control room! I'd want it about 6dB quieter I reckon for most of the day.

Still something I definitely plan on looking into, thanks for the discussion (even if it's not strictly what the OP is asking about).
Old 14th January 2017
  #38
Lives for gear
 

I'm writing and mixing dance music. Ever since I started working at a calibrated level my mixes have improved, I'm much more consistent and I can make decisions with much greater confidence. It makes so much sense to remove the extra variables (fletcher munson curve, ear fatigue, speaker compression etc). I'd really recommend it.

Proper gainstaging with something like VUMT is the final piece of the puzzle...

This article is worth a read: Monitor Wizard | Sound On Sound
Old 14th January 2017
  #39
Lives for gear
 
andy3's Avatar
 

just leave 1/3 of headroom everywhere.
Old 15th January 2017
  #40
Quote:
Originally Posted by psycho_monkey View Post
So I had a few replies from some of my fellow engineers on the group I mentioned.

I'm not going to attribute anything to anyone in particular but they're all working engineers in varying fields, some of whom will be familiar, some not.

As I thought, most of the rock guys were kind of "nah - it's a post thing".

One of the classical/film guys (who I actually knew back from Abbey Road during Lord of the Rings) said it's usual anywhere they mix film or TV music. A couple of the more tech-y pop/rock mixers said they have a calibration level, but don't always stick to it - more somewhere just to come back as a reference.

Others didn't do it at all - and I'd be very surprised if in my parts of the world any studio other than the film places (like Abbey Road) did. It's certainly not something I've ever been asked to do or provide info.
Well... even if people aren't aware they are working on a calibrated system, they probably are. I knew a couple of techs at Larabee and Westlake studios a while ago. All of their rooms were calibrated... but if you were to ask any of the guys that work out of those rooms... they probably had no clue.

Quote:
One thing I did notice - we had a relatively loud (not painful) playback in here, and I did a quite iPhone measurement, and it was around 82dB SPL. That was far too loud to work at all day for me, and I have a big control room! I'd want it about 6dB quieter I reckon for most of the day.

Still something I definitely plan on looking into, thanks for the discussion (even if it's not strictly what the OP is asking about).
YES!!! And now you are starting to see and hear the point of it all. Just because the speakers are calibrated to 85dB SPL (or 82 or 79 or 76, etc whatever) doesn't mean you will actually be listening at that volume.

You noticed that 82dB SPL was very loud for you, right? Now imagine where/how you would set your mic gain if the only way to adjust level was with the mic preamp? You would turn them down, right? And that is the whole point!!! Same thing with mixing... if the mix coming out of the speakers is too loud (at 82dB) and the only thing you can use to adjust levels if the track faders... then you are going to pull all the faders down to -inf... and start bringing them up again to a comfortable listening level for you... the speakers and the studio is still calibrated to -18dBFS = 85dB SPL... but that doesn't mean you are actually listening to everything at 85dB SPL...

That is why I was saying earlier, with my -20dBFS = +4dBu = 85dB SPL calibration my ears will literally start bleeding before I could ever clip a track or clip the mix bus.

My studio is calibrated to that level... but all music coming at the listening position isn't at 85dB SPL. It's probably more likely around 75dB or 72dB or quieter. This forces me to work within the dynamic range of the gear that I use.

Try playing around with it and you'll see what I'm talking about even more. You already have very good recording and mixing skills and a lot of experience so it might not change what you do that much but you will probably notice it a little.

But for those that don't have a lot of experience, it will change how their productions sound for the better and they will be recording and mixing at lower levels in their daws, which means their music will be punchier, clearer, more open, blah, blah, blah.... because they won't be clipping their plugins, mix bus, audio interface, outboard hardware, mic preamps, etc... all those super fast transients that might be getting chopped off will still be there.

then at the end, they can "master" the music using tools specifically designed to bring the level up while maintaining transients.

Anyway... the key to getting good mixes is ultimately gain staging, and using the gear within the normal range of operation it was designed to do. Calibrating the studio from mic preamps all the way through to the speakers ensures that everything is working optimally and the performance of all the gear is linear and that "gain staging" will actually occur naturally. By doing something as simple as turning down the monitoring level... that can throw EVERYTHING else out of whack and create a lot of problems at every stage of a production from recording to mixing to mastering.

Using a ref level to calibrate everything is the best (and really the only) way to ensure everything is working within it's optimal range and ensure repeatable results.

And like Explorer just mentioned, this isn't just for Film/TV people... it will help anyone and everyone. Doesn't matter what style of music you are producing. The level to which you calibrate might be different for different styles... but having a calibration level and sticking to it will ALWAYS help.
Old 15th January 2017
  #41
Lives for gear
 

Quote:
Originally Posted by Etch-A-Sketch View Post
having a calibration level and sticking to it will ALWAYS help.
I don't agree. Sometimes my ears are tired and the reference level in studio X is just too loud. So I turn it down. I'll naturally mix softer because I'm tired, so to balance that the level goes down to match.

Another way around this is to simply have a good reference recording one brings to a studio. If one knows what that sounds like in relation to one's own work then playing that back and setting a reference level you're essentially doing close to the same thing.
Old 15th January 2017
  #42
Yes I can totally see WHY it could be good practise - I just think that fewer studios do it than you think. We certainly didn't have calibration volume points in the London studios I worked at, and I was certainly never informed of any such settings as one of the main in-house assistants.

Missing a trick? Maybe. But if you're a commercial I'd room with a different client every day, then I suppose absolute ref levels are less important - just set the volume consistently throughout the project, and return to it as a default? I do tend to do that, even if it's not a calibration as such - I know if my volume pot is at 10pm and it's not comfortably loud, my gain staging is off, for example...

Quote:
Originally Posted by Etch-A-Sketch View Post
Well... even if people aren't aware they are working on a calibrated system, they probably are. I knew a couple of techs at Larabee and Westlake studios a while ago. All of their rooms were calibrated... but if you were to ask any of the guys that work out of those rooms... they probably had no clue.



YES!!! And now you are starting to see and hear the point of it all. Just because the speakers are calibrated to 85dB SPL (or 82 or 79 or 76, etc whatever) doesn't mean you will actually be listening at that volume.

You noticed that 82dB SPL was very loud for you, right? Now imagine where/how you would set your mic gain if the only way to adjust level was with the mic preamp? You would turn them down, right? And that is the whole point!!! Same thing with mixing... if the mix coming out of the speakers is too loud (at 82dB) and the only thing you can use to adjust levels if the track faders... then you are going to pull all the faders down to -inf... and start bringing them up again to a comfortable listening level for you... the speakers and the studio is still calibrated to -18dBFS = 85dB SPL... but that doesn't mean you are actually listening to everything at 85dB SPL...

That is why I was saying earlier, with my -20dBFS = +4dBu = 85dB SPL calibration my ears will literally start bleeding before I could ever clip a track or clip the mix bus.

My studio is calibrated to that level... but all music coming at the listening position isn't at 85dB SPL. It's probably more likely around 75dB or 72dB or quieter. This forces me to work within the dynamic range of the gear that I use.

Try playing around with it and you'll see what I'm talking about even more. You already have very good recording and mixing skills and a lot of experience so it might not change what you do that much but you will probably notice it a little.

But for those that don't have a lot of experience, it will change how their productions sound for the better and they will be recording and mixing at lower levels in their daws, which means their music will be punchier, clearer, more open, blah, blah, blah.... because they won't be clipping their plugins, mix bus, audio interface, outboard hardware, mic preamps, etc... all those super fast transients that might be getting chopped off will still be there.

then at the end, they can "master" the music using tools specifically designed to bring the level up while maintaining transients.

Anyway... the key to getting good mixes is ultimately gain staging, and using the gear within the normal range of operation it was designed to do. Calibrating the studio from mic preamps all the way through to the speakers ensures that everything is working optimally and the performance of all the gear is linear and that "gain staging" will actually occur naturally. By doing something as simple as turning down the monitoring level... that can throw EVERYTHING else out of whack and create a lot of problems at every stage of a production from recording to mixing to mastering.

Using a ref level to calibrate everything is the best (and really the only) way to ensure everything is working within it's optimal range and ensure repeatable results.

And like Explorer just mentioned, this isn't just for Film/TV people... it will help anyone and everyone. Doesn't matter what style of music you are producing. The level to which you calibrate might be different for different styles... but having a calibration level and sticking to it will ALWAYS help.
Old 15th January 2017
  #43
Quote:
Originally Posted by mattiasnyc View Post
I don't agree. Sometimes my ears are tired and the reference level in studio X is just too loud. So I turn it down. I'll naturally mix softer because I'm tired, so to balance that the level goes down to match.

Another way around this is to simply have a good reference recording one brings to a studio. If one knows what that sounds like in relation to one's own work then playing that back and setting a reference level you're essentially doing close to the same thing.
turning it down from ear fatigue or checking your mix at different levels is not the same abandoning your ref calibration... you still have a ref level calibrated and are using it.

You still started your song at the same ref level that you did your other songs.. and so on... you still jump up and down between your ref level and other listening levels.

It's not like you started one song using -18dBFS = 79dB SPL and then you started a second song using -18dBFS = 68dB SPL and then a third song -18dBFS = 64dB SPL.

when I'm tracking I get sounds using the ref level... and then after we are rolling the first song for a minute or two, I hit the dim button and drop my monitors down by 20dB. Does that mean I'm not using my ref? Absolutely not. The next song we start tracking I'll turn dim off for a minute and make sure that the levels are good for this song and then hit dim again. and so on...

When mixing I start getting a basic balance between all the tracks at the ref level... then i'm constantly jumping around between hitting dim off and on, and moving my volume knob to see how the mix sounds at different levels... but when I decide to make a change... I jump back to the ref level to make the change... hmmm is my kick drum too soft??? Turn the speakers down 30dB SPL, can I still hear the kick? eh, it seems a little buried, maybe it should come up a few dB... jump back up to my ref level... turn the kick up a few dB so I notice it sounds a hair louder at the ref level but still sounds good and musical... then drop back down 30dB... yup, kick doesn't sound buried at a soft volume anymore... and it still sounds good up at the ref level.... and so on...

Is that me not using my ref level? Absolutely not... I'm still using it and refer back to it whenever I'm making major changes.

It's not like the first song I use 85dB SPL as my ref... and then on the second song I never go back to 85dB SPL, I only go as high as 74dB SPL. That isn't a very accurate way to mix if you are doing that... because now the mix you are doing at an overall quieter level than everything else will have more/too much bottom and and probably top end compared to your other mixes for the same project, because your ears hear less of top and bottom at lower volumes.

It's all about consistency. If you use 74dB SPL for an entire album then great, all the tracks on that album will have a consistent amount of bottom and top and have the same relative balance between the tracks. But if you use different overall ref levels from one song to the next you'll start to notice you end up doing more and more tweaks and revisions across the whole album. You'll be constantly going back to fix stuff... and if you use a different ref level when you go back than the ref level you used when you first mixed it... forget about it... the mixes are going to constantly be a moving target.

I work on about 60 to 100 albums of music every year... when moving through so much music so quickly, in my own experiences anyway, i've definitely noticed that using a calibrated ref level really helps keep everything consistent across the board, day in and day out, month to month, year to year.

If people disagree, no harm no foul... to each their own. I'm just sharing my experiences and what I've found helps.

But to those who constantly find themselves having to do revision after revision of the same songs or album and just can't seem to get it right and one day the bass and kick are too quiet, the next day the hats and synth are too quiet, the next day the vocal feels too low, and so on and so on and you are just constantly tweaking your mixes and can never seem to get them right....

or when tracking on one project you are clipping tracks left and right and on the next projects the meters are barely moving and then on the next project everything is clipping again...

then you should try calibrating your setup and studio to a ref level and stick to it day to day... you'll start to see consistency... from day to day, and less rounds of tweaks on final mixes.
Old 15th January 2017
  #44
Lives for gear
 

Quote:
Originally Posted by Etch-A-Sketch View Post
turning it down from ear fatigue or checking your mix at different levels is not the same abandoning your ref calibration... you still have a ref level calibrated and are using it.
If I change the level I'm abandoning the reference, by definition. I'm in TV so we're mandated to hit specific levels. This means I can look at a meter to be technically correct. But the point is that a room calibrated at X where I choose to turn down to -4X is no longer operating at X, correct?

Quote:
Originally Posted by Etch-A-Sketch View Post
It's not like you started one song using -18dBFS = 79dB SPL and then you started a second song using -18dBFS = 68dB SPL and then a third song -18dBFS = 64dB SPL.
Actually, yesterday I mixed two episodes of a show, and I probably had the playback system 3dB down from where the previous episodes were mixed. It makes no difference as long as I understand relative levels, compression levels etc.

Quote:
Originally Posted by Etch-A-Sketch View Post
But to those who constantly find themselves having to do revision after revision of the same songs or album and just can't seem to get it right and one day the bass and kick are too quiet, the next day the hats and synth are too quiet, the next day the vocal feels too low, and so on and so on and you are just constantly tweaking your mixes and can never seem to get them right....
I'm not opposed to calibrating one's playback system. I'm actually in favor of it. I was only addressing what you wrote which I quoted. Perhaps I parsed your statement too narrowly.
Old 15th January 2017
  #45
Lives for gear
 
andy3's Avatar
 

Regard the post processing matching part, I dunno but sometimes If I push out a compressor it gives me a different result than normal volum up post comp via fader.
I started total-noob never matching post processing / then meticolously match all the volume post / now I care on matching outs less (as long as I don't overdo) and I improved, still a niaubs tho'
Old 16th January 2017
  #46
Gear Head
 

A little history, and maybe some understanding of why we care about this... Quick review... Back in the analog days, in pro studios, "0 VU" equated to an output of +4 dBu which is a voltage of 1.228 volts RMS. Note that "0 VU" is a *reference* level which means that this reading is calibrated. "0 VU" doesn't *always* mean +4 dBu. In different environments (older prosumer gear was calibrated to have 0 VU = -10 dBV, or certain broadcast standards had different calibrations as well) it can be calibrated to something else, but for our purposes (studio recording), this is the standard.

Generally pro gear is designed to start clipping around +24 dBu. So if +4 dBu = 0 VU, then that means in analog gear, there is 20 dB of headroom built in above the nominal recording level before clipping occurs. It's worth noting at this point that when say, analog tape starts to saturate, it tends to do so FAR more gracefully than digital, and the same applies to analog circuits as well (although distorting an analog circuit IMO is still generally undesirable but there's always exceptions and there's a million ways to do things, and some gear has become popular because of the way it breaks down so... don't take that as gospel). So, if analog gear starts to clip at +24 dBu and 0 VU is calibrated to +4 dBu then... well, you can perhaps see how trying to tack a label of "0 VU" to -20 dBFS came about.

So now digital comes along. For decades engineers were used to working at a "reference" level. Because of the forgiving nature of analog tape, the occasional clip wasn't a big concern and the properties of tape as it reached its saturation point was often desirable so working at a standard of 0 VU = +4 dBu worked quite well. What should that be for digital though? Well, all you yunguns should be aware that when digital first came on the scene in pro studios, they were 16 bits at best, but often the reality was that the realistic headroom was 12 bits. In the digital world, 1 bit = ~6 dB of headroom so, at 16 bits, you have a best case scenario of ~96 dB of headroom. So, what happens when we go down 20 dB from there? Well, that means we're using about 12.65ish bits of data. I won't get in to the implications of that but suffice it to say we're still having the analog vs. digital debate. Back in the day, resolution was a real concern, especially in a multi-track environment (before I go any further, just for a little dose of reality with regards to how obsessive people get over the importance of gear (not that it isn't, but in such discussions, people often lose track of the big picture), I'll mention that one of the best selling albums ever, "Jagged Little Pill" was recorded on ADATs... and James Taylor's "Hourglass" was tracked with Tascam DA-88's and a Yamaha 02Rw and that album still sounds bloody amazing, but I digress).

Back then though when 16 bits was the best case (and you never actually got 16 bits due to noise floors and other issues, and the same holds true today, nobody is actually getting a full 24 bits of resolution from their 24 bit ADCs due to realities of physics)... that presented problems, thus started the practice of setting levels so you could get AS CLOSE to 0 dBFS as possible in order to avoid sound degradation due to quantization error. Anyone who has done any critical listening knows that, if you take a well designed converter (same model) and make two recordings, one at 16 bits, and one at a higher resolution, there are significant benefits on a myriad of levels. Even going from 16 bits to 18 bits makes a huge difference.

Well, now we live in the world of "24 bits" (quotes just to acknowledge that in terms of real s/n and limitations in the analog domain as well as practical concerns such as manufacturing costs, we don't actually have 24 a full bits of resolution available, but that's a different discussion). For those unfamiliar with the binary language of digital, every time you increase your resolution by 1 bit, you double the number of values you have to represent a signal. So:

1 bit = 2 values
2 bits = 4
3 bits = 8
4 bits = 16
~
8 bits = 256
10 bits = 1024
12.65 bits = 6427
14 bits = 16,384
15 bits = 32768
16 bits = 65536

Now check out what happens when we jump to 18 bits:
18 bits = 262,144
... and then 20 bits...
20 bits = 1,048,576
... and 24 bits ...
24 = 16,777,216

These extra numbers means that quantization error isn't as much of a concern anymore, no more grainy sounding reverb tails, etc.

So how many bits are we working with when we use -20 dBFS as our "nominal" level when working at "24 bits"? Well, first theoretical maximum signal to noise we have at 24 bits is 144.49 dB which is 20 * log10 (2^n - 1) where n is the number of bits. So that means that if we choose -20 dBFS as our reference, we're operating at around 20.68 bits and THAT makes a pretty big difference when compared to older 16 bit systems.

So there's a point behind all of this. I think based on practices from the early digital days which are now often ill-advised, people are still worried about recording lower levels for fear that they somehow aren't getting the best sound possible. Consider this... on a 24 bit system, when your meter is at -48 dBFS, *your signal is still being encoded at 16 bits*!!! At -36 dBFS you're operating at 18ish bits and 18 bits sounds considerably better than 16 bits. You'll hear more of a difference going from 16 bits to 18 bits that you will going from 18 bits to 20 bits, and so on.

The quality of even really cheap ADCs these days is truly amazing (at least if you've had the contrast of hearing the old stuff, people are free to disagree with me on that). So, don't worry about recording as hot as possible. DO watch your peaks. DO leave headroom. You don't need to be shooting for peaks within a few dB of 0dBFS anymore.

Here's another thought I'd like to put in everyone's head... digital gain staging. Think about what happens when you move a fader during mix? If the fader is at 0, nothing happens. If the fader is above or below 0, then math is happening and you're degrading the signal. Now... don't get alarmed. Again, we're working with a 24 bit source here, and the mathematics going on internally is often carried out in 32 bit floating point. If you've moved your faders a little bit, or even a lot a bit it's not a catastrophe that's going to ruin your project. Just think about that though.

With that in mind (and this is the advice I'd give to the original poster), one approach for setting gains in the modern digital world is this... stage your gains so that when you play back your tracks, the rough mix sounds pretty darn good for a starting point with all the "faders" at "unity", and your stereo buss (or whatever format your mix bus is in) is not clipping. This actually isn't a new idea, it's been around forever and if you ever get the opportunity to pull up some tracks that were tracked by a seasoned engineer, you'll notice that if you bring all the faders to unity, you'll already be at a good starting point. It makes for a much better workflow (at least in my opinion) and the mix engineer won't badmouth you. (well, they might... but it won't be over levels).

Last edited by shewhorn; 16th January 2017 at 08:49 PM.. Reason: As usual, didn't finish my thought
Old 16th January 2017
  #47
Lives for gear
 

Quote:
Originally Posted by shewhorn View Post
So there's a point behind all of this. I think based on practices from the early digital days which are now often ill-advised, people are still worried about recording lower levels for fear that they somehow aren't getting the best sound possible. Consider this... on a 24 bit system, when your meter is at -48 dBFS, *your signal is still being encoded at 16 bits*!!! At -36 dBFS you're operating at 18ish bits and 18 bits sounds considerably better than 16 bits. You'll hear more of a difference going from 16 bits to 18 bits that you will going from 18 bits to 20 bits, and so on.
I think what you're implying there is just not correct.
Old 16th January 2017
  #48
Quote:
Originally Posted by shewhorn View Post
<SNIP>
Here's another thought I'd like to put in everyone's head... digital gain staging. Think about what happens when you move a fader during mix? If the fader is at 0, nothing happens. If the fader is above or below 0, then math is happening and you're degrading the signal. Now... don't get alarmed. Again, we're working with a 24 bit source here, and the mathematics going on internally is often carried out in 32 bit floating point. If you've moved your faders a little bit, or even a lot a bit it's not a catastrophe that's going to ruin your project. Just think about that though.
I disagree with you in one small detail - when you move a fader, or do anything else, then "math happens". So the signal is definitely being changed. That's not the same (in my opinion) as saying the signal is being "degraded", with the negative connotations of that word.

I'd argue that the math for digital gain processing is much nearer the ideal "straight wire with gain" that is the holy grail in the analog world. So when you move a DAW fader, ALL you get is gain change, rather than the sort of additional changes that might result in the analog world.

Even back in the days of 14 bits on the PCM-F1, you generally were below the analog noise floor. And analog's gotten a lot quieter, and our digital bit depth has also gone up, so I think it's still not a factor until you do drastic reductions in bit depth.
Old 16th January 2017
  #49
Gear Head
 

Quote:
Originally Posted by mattiasnyc View Post
I think what you're implying there is just not correct.
24 bit max theoretical dynamic range = 20 * log10 (2^24 - 1) = 144.4943
16 bit max theoretical dynamic range = 20 * log10 (2^16 - 1) = 96.3295
144.4943 - 96.3295 = 48.1648 dB

If we're in a 24 bit system then 0 dBFS represents that max theoretical 144.4943 dB signal. Subtract the difference between the two and you end up with a 48 dB difference between 24 bits, and 16 bits, thus, when you're at -48 dBFS on a 24 bit system, that signal is being encoded with 16 bits of precision.

If we did this exercise on a 16 bit signal instead...

16 bit max theoretical dynamic range = 20 * log10 (2^24 - 1) = 96.3295
8 bit max theoretical dynamic range = 20 * log10 (2^16 - 1) = 48.1308
96.3295 - 48.1308 = 48.1987 dB

So, on a 16 bit system, -48 dBFS would be represented by 8 bits of precision, and 8 bits of precision... ahh, the Ensoniq Mirage (doesn't mean you can't do creative things with it). So, going down that far in a 16 bit system, not so great. Going down that far on a 24 bit system... the additional precision afforded to us by 24 bits means it's a lot more forgiving in terms of quantization error to the point where (IMO) it's no longer a concern.

Does that make sense or were you talking about something else?

Last edited by shewhorn; 17th January 2017 at 12:21 AM.. Reason: Thought not finished as usual
Old 17th January 2017
  #50
Gear Head
 

Quote:
Originally Posted by TMetzinger View Post
I disagree with you in one small detail - when you move a fader, or do anything else, then "math happens". So the signal is definitely being changed. That's not the same (in my opinion) as saying the signal is being "degraded", with the negative connotations of that word.
I think that's fair with regards to the negative connotations (I tried to mitigate that a bit but I definitely could have said it more eloquently). There is a loss of precision compared to the original signal for sure but the reality is (in the 24 bit world) I can't think of a situation where anyone would ever be able to hear a difference (well, I can... but nobody would ever do those things... unless they were trying to prove a completely ridiculous point ). I think it was more of a concern in the 16 bit world (thus contributing to this practice of setting gains to the point where every track clipped just shy of 0 dBFS) . With 24 bits though, there's A LOT of wiggle room before you start hearing the effects of quantization error, etc. and much of the math is done in a higher resolution floating point world which further helps to mitigate certain issues present with integer math and errors as such.

Good catch. Didn't mean to insight any fear in anyone, but I know how literal comments can sometimes be taken, especially for those just learning. It can be so difficult to take something like what I said an understand the real world implications. For the folks who are learning... don't be worried about moving your faders in the 24 bit world, the medium is quite forgiving (even saying "forgiving" probably lends itself to too much alarm... just don't worry about moving your faders, period) and you'll never be able to perceive a difference in quality (unless say... you tracked at -60 dBFS... then you might start hearing things :D ).

Last edited by shewhorn; 17th January 2017 at 12:17 AM.. Reason: Thought not finished as usual
Old 17th January 2017
  #51
Lives for gear
 

Quote:
Originally Posted by shewhorn View Post
So, on a 16 bit system, -48 dBFS would be represented by 8 bits of precision, and 8 bits of precision... ahh, the Ensoniq Mirage (doesn't mean you can't do creative things with it). So, going down that far in a 16 bit system, not so great. Going down that far on a 24 bit system... the additional precision afforded to us by 24 bits means it's a lot more forgiving in terms of quantization error.

Does that make sense or were you talking about something else?
But any one given sample isn't a complete signal. The signal emerges upon reconstruction. So any given sample only represents a value, an amplitude, at one given moment in time.

So my point is that the "precision" doesn't really exist (at that point), it is instead dynamic range over time that is of concern. That one sample "down that far" in Xbits per second is still just a value, and you can't represent it more or less accurately throughout that range.

In other words, a sinewave at -10dBFS isn't described with better precision than one at -50dBFS. Our concern instead is that if we end up with "poor conversion" it isn't that representation of that point in time that's the problem (the amplitude of the sample) but that turning it up exposes an increased noise floor.

Or are you proposing that we actually get more precision by recording a signal 'hotter'?

Put differently: I'm not gaining any resolution by describing a value that is quantized to the nearest positive integer just because I place it somewhere else within my allotted four digits...;

0048 has the same "precision" as
3148

Correct? That we're using binary is irrelevant.
Old 17th January 2017
  #52
Gear Head
 

Quote:
Originally Posted by mattiasnyc View Post
0048 has the same "precision" as
3148

Correct?
No. I think you're confusing the time domain with amplitude. If we're representing a sine wave with 16 bits of precision (-48 dBFS on a 24 bit system, or 0 dBFS on a 16 bit system), that's means we have about -32,767 representing the trough of the waveform, and +32,767 representing the peak. At a sampling frequency of 44.1 kHz, that means a single cycle of a 1 kHz sine wave will be represented by 44.1 samples. This is a constant regardless of precision at that sampling rate for that frequency. To accurately represent that sine wave, we need to make sure that we have enough precision so that each time the ADC samples a point of that analog signal, there is a discrete value which it can be represented by, but also, we want enough precision so that when the DAC takes and integer and converts it to a voltage on the other side, it is as close to the original voltage as possible. As you reduce the precision, you reduce the number of steps. Eventually you'll get to a point where two samples will produce the same number, even though the voltage on the input was different. Now you have quantization error.

If you want to do a quick test to hear the effects of reduced precision, just get a tone generator plugin, then follow it by a bit-crusher.

I'm not going to do the actual math here for a sine but you'll get the idea... extending this exercise a bit... 44.1 samples to represent 1 cycle, that means a quarter cycle which represents 0 to the peak of the sine is represented by 11.025 samples. So, the beginning of the waveform will be represented by 0, the 2nd sample will be 2972. Let's say we lob off 8 bits so now you have a situation where we're still representing a single cycle of a 1 kHz sine with 44.1 sample (at a sampling frequency of 44.1 kHz) BUT... instead of having -32,767 to +32,767 we're now limited to -127 to +127. That being the case, sample one will still be 0, but now sample 2 will be 11.5... but wait, we can't represent 11.5 in an integer system so it has to either be 11 or 12. Now you're no longer representing that waveform as accurately as you were when you were using more precision.
Old 17th January 2017
  #53
Gear Head
 

Quote:
Originally Posted by mattiasnyc View Post
...but that turning it up exposes an increased noise floor.
Slightly different discussion than what I was after but yes, bringing the level up of something that was recorded too low will indeed bring up the noise floor. We don't want that.

"Or are you proposing that we actually get more precision by recording a signal 'hotter'?"

I'm saying that back in the days of 16 bits, that happened quite a great deal. It was more of a concern if you were recording music with a lot of dynamics. One of the areas that you could often hear breaking apart if your levels were too low was in reverb tails, or the decay of a cymbal crash. Relative to what the OP was saying... if your average levels are below -18 dBFS... yeah, don't sweat it. Not a big deal. We have so much resolution now, and the analog electronics are so good even on inexpensive gear (lower noise floors) that having your average level represented at -30 dBFS (if that's where you need to record at to avoid digital clipping, I'm suggesting an specific reference level) is not a big deal because even at -30 dBFS you're still recording an average signal with more than 18 bits of precision and as such, it'll still sound great.

Since the OP has run in to some challenges using -18 dBFS as a reference level, I'm advocating for a focus on workflow instead. Track your levels so that when all faders are at "unity", you already have your rough mix, and also, your levels at your mix bus look good (with peaks around whatever kind of headroom you'd like to leave for your mastering engineer). That's a good place to start and experiment (in my opinion at least). Practicing that I'm sure will generate some questions. Maybe the OP will come to the conclusion that maybe they could track most sources with 18 or 20 dB of headroom, but that there's a few things that need more than that. There's tons of solutions to that problem. One would be to make exceptions and record the average level of that instrument a bit lower, or perhaps they'll decide that using a limiter to take a few transients here and there might work better for them. There's tons of approaches and solutions.
Old 17th January 2017
  #54
Quote:
Originally Posted by shewhorn View Post
No. I think you're confusing the time domain with amplitude. If we're representing a sine wave with 16 bits of precision (-48 dBFS on a 24 bit system, or 0 dBFS on a 16 bit system), that's means we have about -32,767 representing the trough of the waveform, and +32,767 representing the peak. At a sampling frequency of 44.1 kHz, that means a single cycle of a 1 kHz sine wave will be represented by 44.1 samples. This is a constant regardless of precision at that sampling rate for that frequency. To accurately represent that sine wave, we need to make sure that we have enough precision so that each time the ADC samples a point of that analog signal, there is a discrete value which it can be represented by, but also, we want enough precision so that when the DAC takes and integer and converts it to a voltage on the other side, it is as close to the original voltage as possible. As you reduce the precision, you reduce the number of steps. Eventually you'll get to a point where two samples will produce the same number, even though the voltage on the input was different. Now you have quantization error.

If you want to do a quick test to hear the effects of reduced precision, just get a tone generator plugin, then follow it by a bit-crusher.

I'm not going to do the actual math here for a sine but you'll get the idea... extending this exercise a bit... 44.1 samples to represent 1 cycle, that means a quarter cycle which represents 0 to the peak of the sine is represented by 11.025 samples. So, the beginning of the waveform will be represented by 0, the 2nd sample will be 2972. Let's say we lob off 8 bits so now you have a situation where we're still representing a single cycle of a 1 kHz sine with 44.1 sample (at a sampling frequency of 44.1 kHz) BUT... instead of having -32,767 to +32,767 we're now limited to -127 to +127. That being the case, sample one will still be 0, but now sample 2 will be 11.5... but wait, we can't represent 11.5 in an integer system so it has to either be 11 or 12. Now you're no longer representing that waveform as accurately as you were when you were using more precision.
This has been discussed to death ad nauseum here on GS in several different threads over the years. Do we really have to hash this out again???

by your account

"No. I think you're confusing the time domain with amplitude."

you are inferring that they are not interconnected. Mattias is stating that they are related to each other.

Case in point... pretty much every ADC and DAC on the market today are actually 1bit Delta Sigma designs at very fast sampling rates. The chipsets used in most converter boxes are from Analog Devices or Texas Instruments and most of them are 1bit (not 16 or 24 or 32)... If what you were saying were true... Then a 1bit ADC or DAC of any type wouldn't work because amplitude is not related to the time domain...it's either on or off regardless of sampling speed.

The reason why they can create a 16/24/32 bit word length is specifically because of the time domain.
Old 17th January 2017
  #55
Gear Head
Interestingly, I just ran across an interview with Chris Lord-Alge earlier today in which he discussed the advantage of mixing at a low level:

(Forward to 1:22)

Old 17th January 2017
  #56
Lives for gear
 

Quote:
Originally Posted by shewhorn View Post
No. I think you're confusing the time domain with amplitude.
No, I don't think I am. We always deal with amplitude regardless of whether we're in the time or frequency domain.

Quote:
Originally Posted by shewhorn View Post
If we're representing a sine wave with 16 bits of precision (-48 dBFS on a 24 bit system, or 0 dBFS on a 16 bit system),
But that's the problem right there. You're not representing a sine wave with 16 bits of precision in a 24 bit system, you're representing a sine wave of amplitude X with 24 bits in a 24 bit system. That it only "fills" a certain amount of bits doesn't matter.

Quote:
Originally Posted by shewhorn View Post
I'm not going to do the actual math here for a sine but you'll get the idea... extending this exercise a bit... 44.1 samples to represent 1 cycle, that means a quarter cycle which represents 0 to the peak of the sine is represented by 11.025 samples. So, the beginning of the waveform will be represented by 0, the 2nd sample will be 2972. Let's say we lob off 8 bits so now you have a situation where we're still representing a single cycle of a 1 kHz sine with 44.1 sample (at a sampling frequency of 44.1 kHz) BUT... instead of having -32,767 to +32,767 we're now limited to -127 to +127. That being the case, sample one will still be 0, but now sample 2 will be 11.5... but wait, we can't represent 11.5 in an integer system so it has to either be 11 or 12. Now you're no longer representing that waveform as accurately as you were when you were using more precision.
I don't see how that's correct. You're obviously going to have values that fall between integers, and you have no higher resolution or greater precision just because the "signal" is louder. You get exactly the same type of error if your signal at sample X is +11.5 as if it is +14258.5. You don't have a decimal point to represent the .5 so you get a rounding error regardless of "where you are on the scale".

Again, the precision or amount of detail or resolution or however you want to call it is the same regardless of whether or not I want to encode the number 4 or the number 678 if my available range is 00000-99999, no decimals. I can do 00004 just as accurately as 00678. Neither gets a decimal point. The next higher or lower value in either case is exactly the same (the LSB).

So, I reiterate: It is my understanding that the issue isn't that we're not getting the precision we want when recording at a low level, but rather that we end up possibly bringing up the noise floor to compensate for the lower level.
Old 17th January 2017
  #57
Gear Head
 

Quote:
Originally Posted by Etch-A-Sketch View Post
you are inferring that they are not interconnected.
They are discreet settings. Resolution, and sample rate. Both affect sound quality, but changing sample rate doesn't somehow change resolution or vice versa. Obviously you can't recreate a waveform without both. That's irrelevant to the topic of levels. Open up the file and what you have is a stream of signed integers. How it got there whether it be PCM or Delta-sigma, and if the actual internal sample rate is oversampled or not, and if dither was used to reduce noise or not... that's all irrelevant to the point.

To reiterate, if you have a sine wave that's peaking at -48 dBFS in a 24 bit system, that sine wave is being represented by 16 bits of precision. If I understood correctly, he was saying that wasn't true, but it is (and I know this because I've written software that writes out multi-channel calibration tones in AIFF format, the WAV format isn't terribly different). Internally you're usually going to a 32 bit float after your DAW reads the file but that's also irrelevant to the OP.

The takeaway from all this is that today's gear is quite capable. That analog side of things is usually pretty quiet, there's gobs of headroom even in cheap AIs, and working with 24 bit audio provides for TONS of headroom such that, if the OP is trying to use -18 dBFS as a reference level, and finds they don't have enough headroom, it's okay to use a lower level because even at say -36 dBFS, you're still working with around 18 bits of precision. I'm not advocating a specific reference level. My suggestion is to try to set gains so that when they bring the recording back up, the rough mix sounds good with all faders set to unity, and the output levels on the mix bus meet whatever standards they have in place. The other option (if they insist on having their reference level be -18 dBFS and fitting everything into 18 dB of headroom) is to use dynamics before the A/D conversion, or to learn the song being recorded and physically ride the gain input on the sections where there are peaks. All that is extra work though (and more expensive in the case of outboard gear) and it won't yield a better result.
Old 17th January 2017
  #58
Gear Head
 

Quote:
Originally Posted by mattiasnyc View Post
But that's the problem right there. You're not representing a sine wave with 16 bits of precision in a 24 bit system, you're representing a sine wave of amplitude X with 24 bits in a 24 bit system. That it only "fills" a certain amount of bits doesn't matter.
Hmmm... perhaps we're dancing around semantics? If you record a sine wave, and the peaks meter at -48 dBFS (on a 24 bit system), I can guarantee you (because I've written the software that does this) that there are 16 bits of precision representing that waveform. If you open up that file, the peaks are going to fall at +32,767, and the troughs at -32,767.

Quote:
I don't see how that's correct. You're obviously going to have values that fall between integers,...
I'm starting to think that if we talked on the phone we'd probably end up agree with one another but given the nature of text... we're both missing a few assumed variables in each others' head... maybe?

Quote:
So, I reiterate: It is my understanding that the issue isn't that we're not getting the precision we want when recording at a low level, but rather that we end up possibly bringing up the noise floor to compensate for the lower level.
Question, when you refer to noise, are you referring to the noise floor in the recording (noise recorded that was inherent in the analog electronics) or digital noise introduced via quantization error?

Either way, if the gains are set such that when you bring the mix up, and all faders are set to unity, you get a good rough mix or starting point with respectable levels on your main bus, I don't think noise is going to be much of a concern (and just to be sure, I'm not advocating a SUPER low reference level like -48 dBFS, I just wanted to make the point that hey... even when you're all the way down there, your audio is still very well represented so recording a little bit lower than -18 dBFS wouldn't be something that would concern me at all).
Old 17th January 2017
  #59
Lives for gear
 

Quote:
Originally Posted by shewhorn View Post
Hmmm... perhaps we're dancing around semantics? If you record a sine wave, and the peaks meter at -48 dBFS (on a 24 bit system), I can guarantee you (because I've written the software that does this) that there are 16 bits of precision representing that waveform. If you open up that file, the peaks are going to fall at +32,767, and the troughs at -32,767.

I'm starting to think that if we talked on the phone we'd probably end up agree with one another but given the nature of text... we're both missing a few assumed variables in each others' head... maybe?
Possibly yes.

My 'objection' is to the use of the term "precision". The way I understand it the precision is inherent in the length of the word that is used, not the magnitude of the value it holds. Therefore, even if your amplitude is -48dBFS in a 24 bit word, your precision is still 24 bits, not 16. It doesn't matter that the first (leftmost) 8 bits are all zeroes.

In fixed point processing your next value is always going to equal the LSB, and that is what determines the actual smallest value / change you can represent.

So by using the word "precision" I think you might be implying to people reading that the entire signal they're recording is somehow less "defined", or has less "resolution", because it's "only using 16 bits out of 24", and I think that's fundamentally false. "Resolution" or "precision" translates to a larger dynamic range, not a better defined signal.

Quote:
Originally Posted by shewhorn View Post
Question, when you refer to noise, are you referring to the noise floor in the recording (noise recorded that was inherent in the analog electronics) or digital noise introduced via quantization error?
The latter.

Quote:
Originally Posted by shewhorn View Post
Either way, if the gains are set such that when you bring the mix up, and all faders are set to unity, you get a good rough mix or starting point with respectable levels on your main bus, I don't think noise is going to be much of a concern (and just to be sure, I'm not advocating a SUPER low reference level like -48 dBFS, I just wanted to make the point that hey... even when you're all the way down there, your audio is still very well represented so recording a little bit lower than -18 dBFS wouldn't be something that would concern me at all).
Sure, we agree. This is all pretty much a digression....
Old 17th January 2017
  #60
Pitying the OP at this point a little - though good info if he can get through it!
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump