"Wasting bits"

[johnS]

A while back, digital recording gurus were urging us to get the signal as hot as possible before hitting the converters, otherwise you're "wasting bits." The danger, of course, is that the hotter you run the signal, the higher the risk of clipping ("running out of bits?").

I gave up on this theory a while ago, and it was nice to stop worrying about overs all the time. Interestingly, my recordings sound just as good, if not better, since I stopped worrying about "wasting bits."

Was this folklore to begin with? Or is there something to it that I'm missing?

[dagosto]

I think this whole "record hot as possible" business goes back to a time when people were recording at 16 bits. It is true that the hotter it is the more dynamic a track can be. When you are recording at 24 though the dynamic range is basically extended pretty deep into the Less significant region. This make raising the volume of a track a few decibels just as effective recording it hotter in 16.

One thing that many recording types don't seem to realize the problem reducing volume with a digital fader. If something is recorded very hot and then hast to be reduced in volume. . . well just think about what is happening to the bits. A 32-bit internal mixing engine which is the standard in most DAWs is helpful to this end.

[johnS]

OK, I give up. What exactly happens to the bits when you slide a fader down in your DAW? I usually never go above unity gain on any track so as to avoid overloading the mix bus. Please don't tell me that's wrong, too!

[dagosto]

Well, when you turn down the fader the bits are not getting smaller, you are using less of them. The more you turn down the fader the less accurate your signal out becomes. Even though 32-bit mixing engines help minimize this effect this is one of the big reasons digital mixes sound so flat compared to analog mixes.

Try this out for yourself. Record something really hot (careful not to clip). Now try turning the fader in your DAW way down and turn your monitor up to bring it to a decent level. If you ignore all the hiss being generated by doing this you will notice that there is basically no dynamic range. It really brings out the "digitalness" of the sound.

[johnS]

I will have to try that. My mixes usually end up with a bit of automation on the channel faders, depending on what's on any particular track, but for overall level changes I usually leave the track at unity gain and do the level change in a plugin, like a waves L1 or a waves EQ. I think these are 32-but internal processing, so perhaps this is less degrading to sound. Digital audio is certainly and odd beast!

[dagosto]

Yeah it is. Most Daws have internal 32 bit mixing because of this so unless you are using an older DAW the channel faders usually end up having the same resolution as the plug in faders.

[Nubus]

I am still lost as to why you would mix using the output of a plug-in as opposed to the fader. what would the advantage be?

[johnS]

One reason I tend to adjust overall levels in plugins is to avoid clipping. I use protools LE, and I often place an L1 limiter as the last plug on a channel. If I need more level, I know I can push it in the limiter, and any stray peaks will be limited rather than clipping the channel. (For this reason, I don't understand why EVERY plugin, channel, and bus in a digital system doesn't include a limiter as the last stage before the output. Why would you ever want to go above 0dBfs?)

I often DO set the "ballpark" level of a track using the channel fader, and I also automate fader moves on the channel itself. But once you do that in Protools LE, you can't adjust the overall level of the track without switching to the track view, scrolling to the end of the song, grabbing the scrubber tool, and making the adjustment. For me, it's usually much easier to stay in the mixer screen and just grab the fader on the L1. This lets me make global level adjustments on the track while keeping the channel fader automation intact. The only way this doesn't work is if you've written automation on the L1, which I never do. (Well, never say never, but it would be a rare case.)

Anyway, just how I've learned to work in ProTools. Any reason why this is a bad practice?

[dagosto]

Originally Posted by johnS

(For this reason, I don't understand why EVERY plugin, channel, and bus in a digital system doesn't include a limiter as the last stage before the output. Why would you ever want to go above 0dBfs?)

Some do. All of the plugs at http://digitalfishphones.com/ do. They are all dynamic processors though that do not take much CPU. I think the main reason most plugs do not is because a good sounding digital limiter is not the easiest thing to design and it would take additional cpu. Also it you are using a plug across a bus the limiter would effect the entire program not just the one instrument. This would make it nearly inmossible to go through your mix and tame peaks in the program by automating volume in tracks that are not peaking.