setting record levels into ProTools with the intent of mixing in ProTools

[Felix]

thanks for those links Mike! those were exactly the threads i vaguely recalled reading a while back. since then, i've been recording lower. reading the recent Digi manual got me thinking again.
so by recording at -18 dbFS i lose 3 bits? and losing those 3 bits is worth it when mixing in ProTools. there is a clash in logic here. why would Digi design ProTools in this way? it doesn't make sense. it is best to record near 0dbFS for the best quality (nearest to 24 bit resolution) yet these optimal recorded signals are not most compatable for mixing.
this is not an irrelevant topic, digi. why won't you respond? the ones who design the system ought to have the best anwser, right?
[img]images/icons/confused.gif[/img]

[music]

I have an 882/20. I am not using a board for input. I am going from pre's directly into the converters. My question is do I use the -10 or +4 settings in the hardware setup? I have been using +4 and it has been working fine.

Just Curious,

Keith [img]images/icons/smile.gif[/img]

[Robert U]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>Originally posted by music:
I have an 882/20. I am not using a board for input. I am going from pre's directly into the converters. My question is do I use the -10 or +4 settings in the hardware setup? I have been using +4 and it has been working fine.

Just Curious,

Keith [img]images/icons/smile.gif[/img]<HR></BLOCKQUOTE>

It all depends what you're taking in . If it's a pre, like you're having, it probably should be +4. Line level, is usually -10 though.

-Robert

[Corey Shay]

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>Originally posted by Felix:
thanks for those links Mike! those were exactly the threads i vaguely recalled reading a while back. since then, i've been recording lower. reading the recent Digi manual got me thinking again.
so by recording at -18 dbFS i lose 3 bits? and losing those 3 bits is worth it when mixing in ProTools. there is a clash in logic here. why would Digi design ProTools in this way? it doesn't make sense. it is best to record near 0dbFS for the best quality (nearest to 24 bit resolution) yet these optimal recorded signals are not most compatable for mixing.
this is not an irrelevant topic, digi. why won't you respond? the ones who design the system ought to have the best anwser, right?
[img]images/icons/confused.gif[/img]<HR></BLOCKQUOTE>

The theory behind recording all individual tracks at the highest level possible goes way back. Lets go back to analog for a second. In any case you probably don't want your hi-hat to be nearly as loud as say your vocal track, but when going to tape it's almost a no-brainer. Sure you could simply record the hi-hat at a low level and in the end mixdown have all faders near unity, but what have you gained by doing that? Most likely noise. Now on the other hand if we recorded the hi-hat at a higher level it will be higher above the noise floor. Then in mixdown when we lower the fader for the hi-hat, the noise goes down with it.

In the digital world, this involves bit resolution, signal to error ratio, as well as signal to noise level just like in analog. But now lets look at a digital mixer like ProTools. If we use up all our bits on the recording of the hi-hat, you should theoretically get a better recording overall. When you adjust a fader to be anywhere but at unity and panned full left or right, you are applying a calculation for that track. But it's not such a complicated calculation. You essentially multiply each sample value by a number that corresponds to its required gain chainge. The beauty behind having a 48-bit mixer though is that the resultant number after the calculation has a bit-rate that is twice as high and twice the precision, and likely preserving the original number within it if another calculation is made on it to bring it back to its original number exactly. So we shouldn't be losing much if at all.

All that being said, however, I can't really say that beyond a shadow of a doubt recording at maximum levels is always the way to go. Sometimes I do actually run into occasions where recording at lower levels sounds better. I'll use vocal whispers as an example. A few times a singer starts whispering on me and before I would make the signer do it again, while I bring the level up. After I do this, well, the whisper starts to sound funny, almost overpresent despite bringing the fader way down. I could possibly attribute this to the mic pres or other amps before the converter altering the sound in ways that are not pleasing but I am far from certain. Also, if I am more conservative with recording levels, I typically run into less problems with plugins overloading (like eq's) than I do when I am on the edge. However with care this ceases to be an issue.

The theory is that to record your levels as high as possible without clipping is a terrific rule of thumb, but might not always apply in every case, but then again, nothing does in recording. There really aren't any rules but your own.

[Greg Malcangi]

<< It's so hard to keep track of all this stuff, but are you saying that if I move a fader in PT, the signal is dithered in the internal PT mixer, but truncated when it goes out through the 888 analog outputs? Even if I use the dithered 24-bit mixer plugin? >>

OK, I'll clarify a bit. The 888 is a 24bit converter so when outputting through the 888 the fader output has to be truncated to 24bit. If you put a dithering plug on the fader the output is dithered before truncation, which is obviously an improvement. However, a greater improvement still is when mixing internally with PT. PT's internal mix buss is 56bit, so the fader output can be calculated at 48bit resolution and the result passed to the mix buss without being truncated. If no truncation is taking place you obviously don't want to be adding dither.

So, if a fader's output is going to the mix buss, use the fader to adjust volume as you would on any mixer. If the fader is being output to your converter, your master fader for example, try to leave the fader at zero and/or dither the output. In light of this you might want to consider setting up a master fader for your external sends and dithering them. I personally think this is a bit of overkill if all you are doing is sending to say an outboard reverb unit.

As far as actual recording levels are concerned, Corey's post hit's the nail squarely on the head.

Greg

[bickele]

I think I'm missing something here...
Does this mean that for every track or group of tracks with a fader level different from 0dB that I want to output to the converters I should create a master fader track with the fader set to 0dB? or I should dither every single track even if routed to the same output?

b.

[Felix]

hmm. two very different very well supported points of view on this topic- two different types of recording level theory. i didn't realize that the master fader was 56 bit. the mixer is 24 bit as stated, correct? btw, greg and corey, thanks for responding in full to the other side of this discusion. [img]images/icons/smile.gif[/img]

[Jules]

We use am Cransong Hedd inserted across the mix buss for 'tape sound'. We split off from its AES output with a Z-sys digital patchbay device to two open record tracks in PT as well as the mix insert..

So we can print a 16 bit dithered (POW-r) mix to DAT or CDR and keep a 24 bit undithered mix for mastering later...

For mixing all in PT I can heartly recomend the new Sony 'Oxford' plug in!

it's really geat..

[img]images/icons/smile.gif[/img]

[Corey Shay]

Here is where Digidesign is not quite clear on its resolution. The mixer is either 48-bit or 56-bit. But lets refer to both of these precision levels as high resolution and 24-bit as low resolution.

This being said, here is my understanding of Digidesign's default mixer plugin. Each low resolution track is passed through a fader which changes its gain and has a resultant high resolution bit depth and sent off to the mixer. All the tracks' high resolution numbers are added together and then sent through the master fader for further gain change. The master fader can recover a clipped mix by shifting bits (its a little complicated). But you can bring the master fader down and prevent clipping by doing this. However for some reason I don't trust the master fader so I always have it at unity and only use it to monitor output levels. Instead I am careful to not clip it by keeping all the individual tracks at a reasonable level. After the high resolution mix is passed through the master fader, the least significant bits are hacked off (truncated) so that the data can be fit through any plugins you have on the master fader and the low res output. Of course the theroy is that truncation is bad, and with the default mixer, there is no way to prevent this from happening. This is why Digidesign has been so gracious and responded to its customers' demand for a dithered mixer (available for download) in that the resultant high resolution mix is dithered back down to lower resultion instead of truncated. The real world benefits to this are inconculsive I think as of now, but I usually use the ditered mixer for peace of mind. Comparing them side by side, though, I can't honestly say I prefer one over the other as I have not had the opportunity to do a real blind test.

Hope this helps some, I know it's tricky stuff. It seems many engineers who have intimate understanding of these concepts (much higher than mine) still don't know the down and dirty specifics of every step of the mixer operation, because no formal documentation on the TDM mixer is available.

[Greg Malcangi]

Hi corey,

<< Here is where Digidesign is not quite clear on its resolution. The mixer is either 48-bit or 56-bit. >>

As I understand it, fader values are calculated at 48bit and passed on to the mix buss without truncation. So the mix buss is 48bits wide. However, there is also an additional 8bit overflow built into the mix buss to provide a little extra headroom when summing 48bit values. So in total the mix buss has 56bit resolution.

<< I can't honestly say I prefer one over the other as I have not had the opportunity to do a real blind test. >>

I did the double blind test between the dithered and un-dithered mixers. There wasn't much of a difference but on some of my mixes I could definitely pick out a higher noise floor with the dithered mixer. For this reason I tend to use the un-dithered mixer.

Bickele:

<< Does this mean that for every track or group of tracks with a fader level different from 0dB that I want to output to the converters I should create a master fader track with the fader set to 0dB? or I should dither every single track even if routed to the same output? >>

If you are outputting a single channel/track with no processing (ie. No pan, no plugs and with the fader at unity) there is no need to dither as no calculations are taking place and there is nothing to truncate. If on the other hand there is some processing going on or the channels you are outputting are a submix for example, you are probably better off creating a master fader, leaving it at unity and dithering it. Having said this, I take it on a case by case basis. For instance I rarely dither the outputs that are being sent to my outboard reverb units. If I'm in any doubt, I set up a master fader with a dither plug and listen out to see if there's much of an improvement. Simple, if you've got a reasonably good pair of ears and a good monitoring environment!

Greg