Recording and Mixing Volum e...

[Blue Bear Sound]

Yeah, mine too. I'm still not sure I see the answer, but it might be implied in there somewhere. I'll have to digest that a bit better...on Tuesday. In the meantime I declare Independance from imaginary numbers and any equations with sigma functions in them.

G.

I don't think there IS an easy answer... waveform arithmetic is not rocket science, but it's not exactly simple either! (Come to think of it, I'll bet there is some waveform theory in rocket science!)

Enjoy your day off!!

 

[Cloneboy Studio]

When he said if you add two -6dBFS signals together, you'll clip. This implies that -6dBFS + -6dBFS > 0dBFS. How is that calculated?

Keep this general rule of thumb in mind:

Whenever you add two signals of equal decibel together the net increase in decibels will double (+6db).

Thus:

-12dbfs + -12dbfs = -6dbfs

If you have two signals hitting -6dbfs when summed you would need to add two more signals of equal level to hit 0dbfs.

Hope that makes sense. Hopefully I got my numbers right.

 

[SouthSIDE Glen]

Keep this general rule of thumb in mind:

Whenever you add two signals of equal decibel together the net increase in decibels will double (+6db).

Thus:

-12dbfs + -12dbfs = -6dbfs

If you have two signals hitting -6dbfs when summed you would need to add two more signals of equal level to hit 0dbfs.

Hope that makes sense. Hopefully I got my numbers right.

That does help some, Thanks . I may be wrong (a distinct possibility at ~11pm on a Sunday night after an afternoon of heavy barbeque), but if that remained linearly true, one would never reach 0dB. For example...

-6dBFS + -6dBFS = -3dBFS
-3dBFS + -3dBFS = -1.5dBFS
-1.5dBFS + -1.5dBFS = -0.75dBFS
etc.

Now maybe my math for doubling is wrong, because that is basically a classic example of an infinite regression that always approaches, but never quite reaches, zero. I must be wrong in my calculation.

Bruce, you may be right that there is no simple answer like I'm looking for, that the math is just too complicated for us to worry our pretty little audio engineer heads about, I know this is already waaaaay off-thread, at the very least.

But something I thought of that may jog someone's brain that's a lot further from diabetic coma than mine is at the moment...maybe conversions to other units would be a shortcut. Change those damn negative dBFS numbers into positive numbers. For example (just of the top of my head as an example), if the numbers were converted to their 16- or 24-bit binary equivalents - with all ones being 0dBFS, obviously - then (I think?) one could just add the binary numbers together and then convert those back to the dbFS value to get an answer.

Still not something easy to do in one's head, but something at least calculable and easy to write a program for. I'd write the program right quick if I had the equation for converting dBFS to binary. But maybe that equation brings us back to imaginary Fourier sigma log thingys again...

I'm going to bed. Have a great and safe 4th all you Americans, and everybody else, enjoy freedom along with us Yankees!

Sorry for the mental rubbish I managed to turn this thread into tonight...

G.

 

[FALKEN]

i think,

dB = 10 log[10] (P1/P2)

i might be totally off there.

but those guys are totally right.

 

[7string]

I have a slightly more complicated view on things.

My starting point for a 24 track, 24 bit recording is -12dbfs RMS, with no greater than -6dbfs peaks (meaning if something is peaking over -6dbfs I will drop the level so that the peaks are hitting there, regardless of RMS). Every 12 additional tracks I will drop the RMS and peak level by -3dbfs to make room on the mix buss.

That way, when it comes mix down time, my faders can be where I expect them to be from my old analog days.

Seems to work well. When I implemented this technique last year the quality of my recordings did go up. Prior to that I stuck with a religious -12dbfs RMS/-6dbfs peak. When I started doing bigger projects (and by golly I love to use 2-3 mics per guitar track and 3-4 takes of guitar... adds up to a lot of tracks quickly) I ran into that digital ceiling.

I am running McDSP Analog Channel on my mix buss these days as well. Helps "soak" any overshoots (that wouldn't clip) and give it that nice analogy flavor (when appropriate).

The only problem for ME is that when I mix to these levels the only way for me to hear it at a decent level is to crank my speakers. Then I forget to turn them back down when I am done and sure enough, a computer event will go off blaring loud and scare the caca outta me...

 

[Chris Shaeffer]

...one would never reach 0dB. For example...

-6dBFS + -6dBFS = -3dBFS
-3dBFS + -3dBFS = -1.5dBFS
-1.5dBFS + -1.5dBFS = -0.75dBFS
etc.

Now maybe my math for doubling is wrong...

Yup. A little off. The key here is that doubling the volume is a 6dB increase. So:

-3dBFS + -3dBFS = +3dBFS... and clipping.
-7 + -7 = -1
-24 + -24 = -18

At least that's my understanding of it.
-C

 

[Chris Shaeffer]

This is why they're called Recording ENGINEER's, by the way.

 

[DonaldChang]

i dont think that's correct....I just di dand expeirement

-24 DB + -24 DBS

It wasnt -12 DB, but -22 DB

 

[FALKEN]

no, its correct. a 6 db increase represents a doubling of volume. your tone must have been at a lower RMS than the peak you read.

 

[Chris Shaeffer]

That's another issue with most DAW meters. They are peak meters-and rightly enough- designed to make SURE you don't clip. Its not a very "musical" way of metering.

More accurate to the way we hear music is the VU style RMS metering that measures more like the average loudness. Its less accurate in that it won't reveal the dangerous peaks but more accurate in that it relects more of what the listener experiences.

I remeber when I realized this and was kind of stunned at how low the RMS meters read for what I thought were really hot signals.

Take care,
Chris