First, if you record at 24bit, there is no reason to record as high as possible. The best practice is to peak each tracks around -16db, because when they add up on the masterbus the mix will be too loud. If you record something and put it against virtual instruments, it's fairly easy to have problems matching the volume if you didn't lowered their outputs.

Second, you need to level match your units, I read that the Boss can have an OUTPUT: -10 dBu/+4 dBu. You should try to match them.

Third, proper gain stage is the best way to avoid problems, you may need to put the Roland gain at maximum and get tons of noises if the output of the Boss is not loud enough. Did you try to adjust the output of the Boss to maximum?

A nice article on gain staging : http://www.soundonsound.com/sos/sep13/articles/level-headed.htm

I belive the -18 dbfs came about as the standard input level, because it's equal to 0db (clipping point) on an analog VU meter.

As far as I understand it, the master bus is actually summing what it's getting from the audio signal post fader. So input settings won't effect the master bus levels.

Sorry bout the italics, it's an iPad thing, doesn't seem to allow me to return to regular txt, or I could be just doing something wrong :)

kmetal, post: 427718, member: 37533 wrote: -18 dbfs came about as the standard input level, because it's equal to 0db (clipping point) on an analog VU meter.

well sorta .... yes -18 (actually -20 to -15, it varies from DAW to DAW) equates approximately to 0dB Vu. 0 dB is not clipping ... it's just a voltage reference. most decent consoles have a ton of "headroom" over 0dB before they actually "clip" (generate square waves). it's not uncommon to shoot for +4 when mixing into an analog bus. on tape for the individual tracks, -3 to +4 is where you want it.

Thanks to everyone for your help. Much appreciated.

Mick

pcrecord, post: 427580, member: 46460 wrote: The best practice is to peak each tracks around -16db, because when they add up on the masterbus the mix will be too loud. If you record something and put it against virtual instruments, it's fairly easy to have problems matching the volume if you didn't lowered their outputs.

Well said - and another nice catch. ;)

It's been my own experience that newer VSTi's - or at least the ones I've worked with - have a tendency to be hot by default - some as hot as -3db by default. The only exception I've ever run into in regard to this, has been with Garritan's GPO and Big Band sample libraries; on earlier versions, the instrument's volume was controlled by the mod wheel on the midi controller, and, because most people keep the mod wheel at a Zero Controller setting when not in use, the output of most of the instruments in those libraries would always be very shy in volume by default.
I never could figure out why they did it that way, it never made any sense to me, but once I figured out that this was how the prog defaulted, I adjusted accordingly.

But... with newer libraries, it seems as though I'm always reducing the output levels of the majority of VSTi's that I use - and as Marco said, particularly when I'm combining them with real tracks, because I generally record audio at around -20 with peaks of -12 or so, particularly since about 9 years ago, when I moved beyond 16 bit and started working with higher bit resolutions like 24, 32 and eventually 32 bit float, after the advantages of working at those higher resolutions was explained to me.

Recently, I opened an ancient Sonar 16 bit project, probably 10 years old, and thought to myself, "WTF ?" when I saw where I had the recording levels on the tracks sitting - they were up around -8db ... and I don't mean peaks, either... I'm talking RMS... LOL :eek:

d.

Kurt Foster, post: 427720, member: 7836 wrote: well sorta .... yes -18 (actually -20 to -15, it varies from DAW to DAW)

It's the converters, not the software, that determines the correlation between full scale and volume units.

So I guess I'm still unclear on input recording leveled effecting the master bus. It's my understanding that that master bus is fed from the faders and auxes, internal matrix. In other words, if I record 24 tracks, all clipping constantly in the red, my bus wouldn't overload, if the faders were low. I'm not trying to be a jerk pcrecord, marco, just trying to clear up my (mis)understanding of the subject.

Lmao, my theory of the lack of headroom on samples, is that, they are selling you a finished product in most cases, and also with vsti s being used live, maybe that's why? Also, how else are you going to get audio bricks on every track? You mean add another 18db of GR on the way in.?

kmetal, post: 427781, member: 37533 wrote: In other words, if I record 24 tracks, all clipping constantly in the red, my bus wouldn't overload, if the faders were low. I'm not trying to be a jerk [[url=http://[/URL]="http://recording.or…"]pcrecord[/]="http://recording.or…"]pcrecord[/], marco, just trying to clear up my (mis)understanding of the subject.

That's exactly it ! You could have as many tracks as you want peaking at -2db and can lower all the track faders or track gain and be safe. Thing is , it's hard to do automation when all the faders are down and you don't have much volume left to play with. So I recommand recording at lower levels or using the Track gains. of course Track gains adds one more processing the computer has to do.. which I try to avoid.

i'm not sure that's how it works. what i understand is the amount of math the processor has to do affects the quality. so in that light, it would be best to record at levels that allow you to set the channel faders at 0 gain / attenuation, or as close as possible so each track plays back with as little number crunching as possible.. ironically, this same approach is usually what worked best (for me) with tape & console. the best mixes always seemed to be the ones where the faders were all set pretty close to nominal setting (in the shaded area).

Kurt, you just better phrase my thoughts, thanks. I was leaving work when I wrote my posts and I didn't explain why I was trying to avoid playing with the gains.
I'm just wandering if anyone as taken time to test this theory that the more CPU calculations you do the more errors can happen which can lead to audio degradation.
I guess it is specially important when doing round trips or hybrids or mixing to a second DAW because all the calculations are made in real time.
I'm also curious as if those risks are still present (or worse) when I do an offline export to a wave file.

I'd be surprised if the effect of running your analog signal at optimum levels or not before it hit the converter didn't completely swamp any effects of processing to adjust gain.

bouldersound, post: 427819, member: 38959 wrote: I'd be surprised if the effect of running your analog signal at optimum levels or not before it hit the converter didn't completely swamp any effects of processing to adjust gain.

I'm not sure I get what you are saying.

If your record levels are off your analog input gain staging is off. That will have more effect than a gain adjustment in the DAW.

bouldersound, post: 427824, member: 38959 wrote: If your record levels are off your analog input gain staging is off. That will have more effect than a gain adjustment in the DAW.

That's for sure ! ;) We are talking to record at -3db or -16, I wouldn't be affraid to create major problems...
Also, recording at a slightly lower levels can also be a good thing for those who have average quality preamps. They rarely take it well when pushed..