Setting levels

  • Thread starter Thread starter KevinDrummer
  • Start date Start date
K

KevinDrummer

New member
I would think that you would want to have the "hottest" signal at the earliest point - but I'm unsure.

What's the proper way to set levels? Here's a sample signal chain...

vocals --> into run of the mill dynamic mic --> into Behrinhger mixer --> into standard PC sound card --> into Cakewalk

I can adjust the trim on the mic pre on the Behringer, adjust the level (channel or main) on the Behringer, adjust my "line in" level on my PC, or the trim in Cakewalk (not sure what that is really doing). What is the procedure to get the best cleanest signal? Would it change if I running a drum overhead? When I arm a Cakewalk channel, I get a flicker on the meter with no signal (at about -66 to -48 db). I'm looking to clean it up. Help. Thanks!
 
Most gear, even not so great gear like that Behringer, has 'sweet spots' where it sounds best. Try to find those.

A good rule of thumb is 70%.

In general if something is set HIGH and something LOW there is something *WRONG* with your setup. Nothing should be cranked or turned down most of the way.

Efficient gain structure is key to a clear recording with maximum headroom, no overs, and optimized signal-to-noise ratio.

I could get more indepth but dinner is ready. :)
 
IMHO, and in many other not-so-H opinions, there are reasons to never record OR MIX with peaks hotter than -6dBfs. At least in 24-bit.

That goes for mastering too, but it'll never happen... Distortion that occurs at the D-A is the culprit. It's a long, long story...

Long story short, without the physics lesson that I'm not qualified to give anyway - Keep your peaks at or below -6dBfs for everything. From tracking until the final mix. Only the final master should go over (actually, it shouldn't either, but the Industry won't stand for that I'm certain).
 
My personal rule of thumb is -15 to -12 dbfs RMS levels with peaks no more than -6 dbfs.

If you have all or a majority of your signal over -6 dbfs the summing will be compromised when you mix down. You will be forced to push some thing down in level and lose some good SNR. Also, you don't want digital summing 'overs' because adding together tracks is exponential.

I would say that over 24 tracks I'd aim for a -18 to -15 dbfs with peaks around -8 dbfs or lower. The more you have to merge, the bigger the chance you may end up with junk.

Digital mixdowns CAN be extremely clean, just like digital recording, if you are careful about gain staging and know what you are doing.
 
As far as gain staging your board goes. Set the fader on the channel and the output (if you are using it) at unity. There is normally a scale on the faders, unity is the zero mark. Now set the trim on the channel so that the average signal level is about -12 or so in the daw. The soundcard should be set at unity as well. This will get you the best signal to noise ratio you can get.
The standard soundcard is most likely giving you the noise problem.
 
Massive Master said:
IMHO, and in many other not-so-H opinions, there are reasons to never record OR MIX with peaks hotter than -6dBfs. At least in 24-bit.

That goes for mastering too, but it'll never happen... Distortion that occurs at the D-A is the culprit. It's a long, long story...

Long story short, without the physics lesson that I'm not qualified to give anyway - Keep your peaks at or below -6dBfs for everything. From tracking until the final mix. Only the final master should go over (actually, it shouldn't either, but the Industry won't stand for that I'm certain).
Click to expand...

I 200% agree with this! At -6dBfs peaks, you STILL have 23 bit resolution capabilities, which is still around 3 or 4 bits more than what your ear can percieve, and quantization errors later on via DSP will still have plenty of room to not be heard.

No need to push the meters hard these days in this 24 bit world. I often set gain so that I can have all the channel faders at Unity gain and all the instruments are basically "mixed" somewhat well. You would be surprised how much better your recordings will come out if you strive for this.

The above works NOT because you will do less DSP later to the audio (well, not directly because of that), it works more because it will force you to make the source sounds fit the overall production better. When you got a bunch of whacked out gain structure, some stuff maybe be FAR too loud, and some stuff far too quiet. You find out later that adjust the volume on these parts unveils that the part just doesn't sound right with the other parts.

Anyway.....
 
Actually, Ed, didn't you post a link to a paper somewhere (probably one of Aldrich's) about the "physics lesson" that I didn't want to get into? You know, the whole "curve of the peak reaches far beyond 0dBfs" at the D-A...

It'd probably be interesting reading for some around here... If it wasn't you, well... well, then I'm obviously mistaken. :o

I'll try to dig around and see if I can find it anyway...
 
Yeah, I posted a link to the "white paper" over at the "other" BBS (sorry, won't name it here...Dragon might BAN ME! :)) But it was actually pipelineaudio that linked me to that article. So, he really get's the credit around these parts for finding it. :)

You have to become a "member" of this website to view the article:

http://www.tllabs.com/index.php?option=content&task=view&id=37
 
Wow I learned something tonight... :)

I always set everything up trying to get a zero peak on everything and ending up limiting it on the master to -6.

Thanks guys i had been wondering about that too!
 
You must log in or register to reply here.
Back
Top