track at lower levels !...some info.

[manning1]

http://recforums.prosoundweb.com/index.php/t/4918/6490/

this thread is a very interesting read about tracking at lower levels.
so i started recording tracks in the daw peaking at -12.
and it seems there is lots of merit to what these industry heavies talk about.
some even suggest recording in a daw at lower than -12 peaks.

anyone got any opinions ??

[HomoNeophilus]

That's a lot to read through... But yeah, I've snapped this up also some time ago and I've changed my tracking behavior since. You get a few nice benefits and really no drawbacks (at least if tracking at 24bits with, theoretically, 144dB of dynamics).
Major benefit being no risk of ever clipping. Another is that soundcards/ADs, especially cheaper ones, usually gets less linear at levels reaching 0dBfs so staying below -6 is recommended. Third is easier mixing without having to turn all track levels down a lot to ensure the output is not clipping. I try to generally record at 15-12dB with the occasional peak at about -6.

[jussi]

Dont forget that your theoretical noise floor might be huge but the pratical one might be around -80db... so the lower the level you record at the more you'll have the real noise floor close to your signal.. just dont go mad and start recording with peaks at -60db or instead of a loudness war we will have a "quiet and hissy" war lol

[Measuring Man]

The best method I've found to maximize your converters quality is to determine the relative 0dbv/u of your sound cards converters and use that as a reference point for interfacing with the analog world, both in and out.

For instance, the 002 I use at one studio has a maximum output of +18dbu, easily found on the spec sheet. That means that if I'm referencing any professional gear that uses the standard +4dbu reference, in the computer I use -14 as my reference point, 18-4 = 14db of headroom. So if I'm recording, my analog meters on my pre-amp are reading around 0dbu, and my audio in the computer is averaging -14db. That's average, not peak. In practice, in means my peaks are usually around -6db in the computer, but that can vary depending on what I'm recording. (if you're referencing -10dbV pro-sumer gear ad aprox. 12 to all dbu measurements to get the equivalent -10dbV reference, unless the gear specifies the dbV scale.)

Likewise, if I'm mixing with an average level of -14 in the computer, when I layback to my analog 1/2" 2-track, which coincidentally has about 14db of headroom over +4du, my reference levels will be the same.

In my pro Tools HD rig, the converters are calibrated to be -20db digital to +4dbu analog, that's down 2 db from their stock setting of -18db digital for +4dbu. This is a broadcast standard, -18db digital in Europe. So I have to mix at -20 in the computer and make sure nothing usually goes above -8 (a broadcast maximum reference by the way, found that out the hard way when tapes got sent back) in order to avoid sending analog tape in it's distortion ranges. Other analog gear has much greater headroom, but like your converters it's best not to push it.

I often use tones, just like they used to in the analog days, and still do in post, to check my calibration.

I find if I use the reference levels, my analog outboard gear, and my converters perform at their max. The cheap power supplies in "pro-sumer" gear usually crap out at higher range of their voltages, so the top 3-6db of you converters range can sound strained or otherwise crappier. 6db in digital translates to about 1 bit of resolution loss, but with noise floors, you're only getting 18-20db range to begin with. But 20 good bits is better than 24 crappy ones.

With the advent of peak metering in digital, the art of understanding db's is going the way of the do-do. Which is a shame, 'cause after I finally figured out how db's work, and it took a looooong time, it's really saved my ass from a lot of digital and analog audio problems, distortion being the biggest problem.

I listen back to the ADAT recordings of the 90's where people were told to slam the levels to maximize resolution, and they sound strained and thin. But any tape that properly referenced the ADAt's relative analog point, -15db digital on the original ADAT's is +4dbu on the balanced out and -10dbV on the unbalanced out, sounds amazing, even though it's only 16 bits. Same for many DA-88 tapes I work with.

[Art Evans]

So I have to mix at -20 in the computer and make sure nothing usually goes above -8 (a broadcast maximum reference by the way, found that out the hard way when tapes got sent back)

I've never understood that - how do they cope with commercial CDs which sit on zero (full scale) most of the time?

[LOSER]

Originally Posted by Art Evans

I've never understood that - how do they cope with commercial CDs which sit on zero (full scale) most of the time?

He is refering to dBu (and/or dBv which is the same as dBu) which is a reference relative to the 0.775V RMS voltage of an alternating current systems.

On the analog tape he is refering to the VU-Meter (which is basicly a RMS windowing of the signal with a 300ms window size) value of -8dB (with tape having a saturation point at around -6dB).

All this is AFAIK, so don't quote me on this one .

[Measuring Man]

When I work in post, and I need to mix in music so it's at "broadcast levels" I mix it in, ie. turn it down, so it's average is around -20. With the way commercial mixes are, there might be as little as 6db in peaks over that, so it might only peak at -14 to -10 usually. However I often deal with music that has not been squished in mastering.

It's all about reading VU/RMS meters as opposed to Peak. Oh , and your ears.

[Staccato]

For most soundcards, 0VU translates to -18dBFS. What this means to me, is that if my preamp is at 0VU and is operating in the proper range, I expect my level in the DAW to be -18dBFS. -12dBFS, IMO, is a perfect target.

[Measuring Man]

Most PRO soundcards yes. The pro-sumer, and USB stuff is usally lower, -12ish digital, due to the fact that the maximum voltage on the line is 5V. Plus, most of those boxes are a fixed reference, so you determine what it is and work with it in order to keep things optimal.

[Youn]

i've been hitting between -18 and -12dBFS for the past ten years or so, ussually mix to that level as well...

[yep]

What a totally awesome thread. Thanks for sharing, manning1.

New readers: be aware that the good stuff doesn't really start 'till Paul Frindle's long post on page 5, which is the turning point in a very long thread that ends up being not at all what it seems on the first page.

For the lazy, the grossly oversimplified version is that digital mixes might be dramatically improved by keeping peak levels low, and that errors related to overly hot digital signals causing clipping that is "invisible" on conventional digital meters might be to blame for much of the "analog summing is better" experience out there.

Cheers.

[geoffroy]

great thread. I'm always lowering the gains of each channels when mixing and it's sometimes a problem with some plugins (ex: compressors).
But how do you cope with that :
- if the soundfile peak at -12db, the waveform is tiny and it's usually not as easy to see it on the screen
- try arming an enveloppe in reaper, and lowering the gain enveloppe. This minimum I can have is -44.6 dB before it reaches -infinite, and it's more difficult to draw the gain enveloppe at these levels

any hint ?

[Youn]

In the arrange window... simply hold down "shift" then press the "up" arrow 14 times... this will "zoom" the peaks virtically so a -12dB peak should look to hit just below the boundries of the audio object...

[geoffroy]

hey! I didn't know that :=) thanks!