recording at -12 db and normalizing to -0.5 db: how much quality is lost?

[trancenonZENsedance]

Also, on page 37 of the mixer's manual where it says "Digital Architecture Specification", it states "Analogue/Digital Line-up: +12dBu=0dBFS". Don't worry, your equipment is all fine and working according to spec.

 

I still don't really understand this ("+12dBu=0dBFS"), but since the manual says it's supposed to be like that and normalizing apparently isn't an issue at all, things are fine for me right now the way they are. Maybe I'll dive deeper into decibels and digital signals at some point and be able to make sense of it, but right now I want to record some mixes, since Audacity just started recording again, much in the same way it did last time when it just wouldn't start recording.

 

Thank you, guys, for all the valuable input, it was of great help to me!

 

aliendna99: The screenshot is Audacity displaying the audio in waveform (the same way a sinus wave is usually depicted going from left to right along and up and down through a straight horizontal line). If you'd still like to see a zoomed in screenshot of the waveform, let me know and how much you'd like it to be zoomed in.

[aliendna99]

No. I know it is a wave file.

What I mean is the picture doesnt make sense to me.

It should go to a maximum ceiling of 0db. http://bayimg.com/eAohIaAEe

 

Your screenshot shows something wierd in my opinion. The above picture is a sample of a track with very little dynamic left, and is probably very loud, with estimate of about -7db average rms.

 

Unmastered it should look entirely different, something like this http://bayimg.com/EAohlAaee

 

as you can see there is plenty of dynamics left. These are just pictures i googled, because I am not home and could not provide any more accurate pictures atm, but I believe you get the idea.

 

 

so I hope you know realise what I meant when I said I was confused by your screenshot.

 

 

 

 

also, -13,8db rms seems a little bit too low imho. If played on a radio it will be very noticably lower in volume comparably

[aliendna99]

and yes, to hopefully end the thread

 

go ahead and peak normalize to 0, its fine and will not decrease in quality

[Djuna]

also, -13,8db rms seems a little bit too low imho. If played on a radio it will be very noticably lower in volume comparably

 

But for own use it doesn't matter.

[aliendna99]

 

But for own use it doesn't matter.

of course not!

 

it was just a piece of information.

[Djuna]

of course not!

 

it was just a piece of information.

 

Okay, I was afraid that you'd suggest to limit the shit out of it.

[Veracohr]

I still don't really understand this ("+12dBu=0dBFS")

Remember what I said about decibels being a relative measurement. There are a variety of common "dB" measurements that are all referenced to different things. "dBu" is referenced to SQRT(0.6V) RMS, while dBV is referenced to 1V RMS, dB-SPL is a sound pressure measurement referenced to 20 micropascals, dBm is a power measurement referenced to 1mW, so on and so forth.

 

dBFS is a digital signal measurement referenced to all bits being 1. dBu is an electrical measurement. Any analog to digital converter will have some design reference that says "this level of analog input voltage results in 0dBFS, or all bits being 1", or vice versa for a digital to analog converter. Your mixer is designed so that 12dBu (which is about 3.08V RMS) is the amplitude of analog voltage that will result from a sample that is all 1's: which is called 0dBFS.

[trancenonZENsedance]

Your mixer is designed so that 12dBu (which is about 3.08V RMS) is the amplitude of analog voltage that will result from a sample that is all 1's: which is called 0dBFS.

 

But where does this come into play when my signal path for recording is entirely digital?

[Veracohr]

Oh, you're recording digitally into the computer? Sorry, I guess I completely overlooked that part.

 

You say the signal is 'about' -12dB. Meaning the average value is about -12dB, but it goes up and down around that point? In that case, that should mean the average amplitude of the music is -12dB, which is fairly normal. You might try taking two different CD's, one which is noticeably quieter than the other, and see if they are indeed going in at different average levels.

 

The average amplitude of music depends on how it is mastered. Mastering engineers will normalize the signal so that the highest peak is at or just below 0dBFS, but the average level depends on the music and how compressed it is. Because of this, normalizing various pieces of music so that all their highest peaks are at the same level doesn't necessarily mean they will sound like they're at the same level. The relative loudness we perceive is due to the relative average level, not the peaks. Or in other words, use your ears, not your screen.

[trancenonZENsedance]

Oh, you're recording digitally into the computer? Sorry, I guess I completely overlooked that part.

 

You say the signal is 'about' -12dB. Meaning the average value is about -12dB, but it goes up and down around that point?

 

No, I want to say that when the mixer displays 0dB on the meters (both, channel and master/mix meter always display the same level when the channel fader is in top position) Audacity records a signal at -12dB and this is what I'm trying to make sense of.

[aliendna99]

It is most likely your mixer then. The 0 on your mixer is probably more like -12 then. In any case just peak normalize and be happy

 

OP,

[trancenonZENsedance]

It is most likely your mixer then. The 0 on your mixer is probably more like -12 then. In any case just peak normalize and be happy

 

OP,

 

I am happy, but this doesn't keep me from wondering why the mixer was designed this way.

[trancenonZENsedance]

http://bayimg.com/eAohIaAEe

The above picture is a sample of a track with very little dynamic left, and is probably very loud, with estimate of about -7db average rms.

 

But right now I don't really care about average rms, since I'm assuming that as long as I get the peaks of the individual tracks close to 0db, the average rms would just depend on the average rms of the individual tracks. I want the peaks to be close to 0db, that is all.

 

so I hope you know realise what I meant when I said I was confused by your screenshot.

 

I'm not sure but I guess it's that the y-axis in Audacity is labelled differently than you're used to?

 

also, -13,8db rms seems a little bit too low imho. If played on a radio it will be very noticably lower in volume comparably

 

So I guess there is something abnormal. Since the track has hardly any breaks, let alone long ones, shouldn't the average dB be much closer to 0dB than -13,8dB after normalizing to -0.01dB?

[trancenonZENsedance]

So I guess there is something abnormal. Since the track has hardly any breaks, let alone long ones, shouldn't the average dB be much closer to 0dB than -13,8dB after normalizing to -0.01dB?

 

I just tried something: Contrast analysis of the left channel of the original track (Leftism - Mind Doctor) returned -9.6 dB. It's not much of a difference, but why is there any significant difference at all?

[Djuna]

 

I am happy, but this doesn't keep me from wondering why the mixer was designed this way.

 

As said before, a dB is always relative to something, so there's no such thing as 'a dB meter'. dBFS =/= dB SPL =/= Vu =/= ... There's nothing wrong with your mixer, it just has different meters than your DAW. The meter in your DAW is a dBFS meter, which means that a signal will clip if it goes over the threshold of 0 dB. However, with your analog mixer you can see that there is more headroom: after 0dB you have numbers going up to, for example, 14dB. Compare it with measuring in inches and centimetres. 10 centimetres = 3.9 inch, that doesn't mean that something is 'smaller' when measured in inches.

[trancenonZENsedance]

 

As said before, a dB is always relative to something, so there's no such thing as 'a dB meter'. dBFS =/= dB SPL =/= Vu =/= ... There's nothing wrong with your mixer, it just has different meters than your DAW.

 

Yes and it does make sense to me that a digital signal is measured in -ndB to 0dB, witt 0dB being the loudest possible sound, like in Audacity. But it still doesn't make sense to me that a digital dj-mixer uses -12 dB as a reference instead of 0 dB for its meters.

[Veracohr]

I'll try another explanation now that I understand the picture.

Notice how those meters have both positive and negative dB numbers? In a digital signal, it's impossible to go over 0dBFS, because 0dBFS is the absolute maximum. But those meters are analog, and are calibrated to the analog part of the mixer, which uses the common dBu reference, which is analog and thus has no absolute maximum. Since the mixer is calibrated such that +12dBu=0dBFS, that means an analog signal which is 0dBu (like you're seeing on the mixer's meters) is equal to

(+12-12)dBu = (0-12)dBFS
0dBu = -12dBFS

like you're seeing on the DAW's meters.

Take special note of the fact that I have used dBu and dBFS, never just "dB". This is because a decibel is a generic relative measurement which can have any number of reference points, but must have a reference point (stated or implied) in order to have meaning. You could even make up your own reference point and calculate signal levels in dB relative to it. But it is important you don't think of -12dBu and 0dBFS as being on the same scale. They are different scales; Djuna's centimeters and inches comment was good. You can have 1/4 of an inch and 1/4 of a centimeter; even though the "1/4" relationship is common to both, they are not the same measurement.

[trancenonZENsedance]

But those meters are analog, and are calibrated to the analog part of the mixer, which uses the common dBu reference, which is analog and thus has no absolute maximum. Since the mixer is calibrated such that +12dBu=0dBFS, that means an analog signal which is 0dBu (like you're seeing on the mixer's meters) is equal to

(+12-12)dBu = (0-12)dBFS
0dBu = -12dBFS

like you're seeing on the DAW's meters.

 

Ok, I get it. I'd have preferred the meters to be calibrated for digital recording, but apparently A&H saw their main purpose in monitoring the analog output.

 

 

So then, I think, there's only this question left for me: When I burn a wav-file as an audio CD, and record it from the CDj-400, through the Xone:DB2, to the PC per Audacity (all digitally connected) and then normalize to -0,01dBu, why does the result only have -13,8dBu average RMS, when the original wav-file had -9,6dBu average RMS?

[Veracohr]

Good question, I don't have an answer for that. Your first post talked a lot about bit depths; maybe between all the bit depth conversions something gets lost? I don't know.