Does DAW have infinite headroom?

[cakeshoppe]

This is only the case in floating point DAWs, Pro Tools for example is 48bit fixed point, and has a fixed amount of headroom (54dB?). And I think the "bigger/louder" sound is an illusion of the composition, rather than being caused by the gainstaging (ie before the clipped portion, the song is still building up). I think there also may be some sort of noisefloor cost to the "infinite headroom" in floating point DAWs but I really don't know enough about the math to say for sure.

[hasbeen]

If you delve far enough in to this thread there is a lot of valuable information regarding this. Good Luck!

The Reason Most ITB mixes don’t Sound as good as Analog mixes (restored)

[mrc]

this is a short article regarding the subject
Q. What's the dynamic range of the channels in my DAW?

[Tom Higgins]

Originally Posted by hasbeen

If you delve far enough in to this thread there is a lot of valuable information regarding this. Good Luck!

The Reason Most ITB mixes don’t Sound as good as Analog mixes (restored)

[Treva]

My cubase pan law is set to EQUAL POWER.
I really don't know what that means.
Didn't even know about that thing.
Should I keep it that way or set to 0?
Could someone gimme a short explanation about it?
Thanks.

[norbury brook]

just google pan law.

MC

[Treva]

Originally Posted by norbury brook

just google pan law.

MC

DUH!
There'll be a gazillion results.
I prefer trusting people here.

[KevWind]

Originally Posted by shambles76

HI GS<

Do check this link out:

DAW Headroom Test - YouTube

I stumble upon this link on youtube. Am a bit baffle Could anyone here help to explain if all this is true? Does DAW really has infinite headroom?
For me the tracks sounded much better at the end of the video.
To me is sounds like a good way to make your tracks sound louder and bigger!

Your time & advise is much appreciated tqvm

The answer to your question is yes BUT

The But is as someone pointed out the DAW must be floating point.
Also as some one stated Pro Tools TDM was fixed. However PT 10 now has 32 float and PT LE has been 32 bit floating since version 8
And now PT HDnative and AXX are floating point capable

Also understand this is just referring to within the software itself. This does not mean that within the entire process there is infinite headroom without clipping. Digital recording and playback definitely has a headroom limit and that is at 0 but this limit is a function of the D/A converter.

here is a great quick explanation that was PM'd to me about this.

"Anyway, the master bus never clips, especially in 32-bit float DAWs. So why does the master fader go red like it's clipping? The DAW is just telling you that you've reach 0 dBFS on the master bus. If it's a 16-bit session, it is telling you that you've used up all 16 bits; if it's a 24-bit session, you've used 24 bits.

Even though the DAW has pretty much infinite headroom internally, it has to send the data as 16-bit or 24-bit chunks to the interface's driver. The driver, in turn, has to send the chunks to your interface's DAC, which has to convert the number into a voltage. The DAC is calibrated such that all-1s (i.e., 24 bits on) equals the highest voltage that its power supply can provide.

Thus the red clip indication on the master fader is really the DAW telling you that the DAC is clipping, not the DAW itself."

[sparqee]

Originally Posted by Treva

DUH!
There'll be a gazillion results.
I prefer trusting people here.

I think you'll find that a lot of people here will respect you more if you do the research yourself (rather than asking to have it handed to you) and then ask for people's opinions on when, how, why to make certain choices. Rather than Google I would actually suggest the Cubase operation manual (page 121 explains the available "pan law" settings). The question of which to choose is a personal one. I prefer equal power most of the time but if I'm automating big special effect pans I will sometimes choose other laws. I just use my ears and decide which I like the sound of.

[theblue1]

Originally Posted by Treva

DUH!
There'll be a gazillion results.
I prefer trusting people here.

Pan law is just something they put in to confuse the rubes.



(That's not true. But it is why you should take a little more initiative and responsibility for educating yourself. If you just 'trust people' on a 'net BB, you're likely to get your head filled with a truckload of nonsense. )


_______________


And speaking of a truckload of nonsense, here are my additional thoughts on FP and DAWs...

With regard to the OP question, using a fixed point system affords the maximum mathematical resolution for the designated number of bits used to store the number -- but, as correctly noted, it is a fixed range.

If you use a floating point system to store such a value, you devote some of that memory to store the 'significant' value, and the rest to store exponent information. It's the exponent information that allows us to 'scale' over a wide range of potential values -- but the precision of the number, whether it's .12345 or 123.45 remains the same.



Again, it's very important to understand that the precision is not unlimited. Think of the precision as a range that can be scaled to much higher or lower values.

So, if you devote 24 bits of memory to store an value, you have a maximum theoretical resolution of 144 dB set to a fixed range. A 32 bit FP system would deliver about the same 144 dB of mathematical resolution but it could be floated so that the maximum value was -20 dB, -200 dB or 20 or 200 DB. But the minimum possible value would still be ~144 dB below the maximum (assuming 24 bits of effective precision).


FWIW, the way FP is used in audio DAWs, a 32 bit FP number can store approximately 25 bits of significant data, with the rest used for exponential scaling. (24 bits for the number, AIRI, and a bit devoted to signifying + or -, since we want to allow values above 0 dB for our DAW purposes.) Common 64 bit FP systems will typically deliver 53 bits, IIRC. (IOW, don't be fooled by the 'coincidence' of a 32 bit FP number providing 24 bits. It's not 'proportional' in a linear sense.)

http://en.wikipedia.org/wiki/Floating_point