Gain Staging

[Toki]

This is really a question about whether I'm understanding gain staging, or rather whether my approach to setting channels levels is right.

I've read a fair bit in the past on the topic and since I'm revisiting writing again now and still trying to refresh my knowledge, I've read through the forum a bit too. And I think I'm on track...

So generally any new instrument I introduce, lets say it's a synth lead, I'll play around with the synth settings, sculpt the sound (rarely do I touch the master out on the instrument), write a melody, play with effects and once I'm happy I have the sound I want, I'll head to the SSL. I'll make some EQ adjustments and then I set the gain knob to hit to -10db on the channel level. -10db I picked up somewhere (I'm guessing this is peak on Reason's mixer channels as switching the master fader to VU does nothing to the channel faders) and I believe I read this is around -18db VU, another figure I hear banded around? I'm also seeing -12db thrown around as a Reason specific thing. Either way, then I'll give the channel fade a nudge for a quick balance in the mix while I work, making adjustments to the gain and fader again if I do anything dramatic to that instrument such as adding new effects.

So the question really is, is this a good approach, I'm I on the right track to setting decent channel levels with this approach? I know everyone has their own little ways of doing things, and we can insert gain meters here and there through the signal chain, and that I'm trying to boost my signal, reduce my noise and end up with a nice set of balanced channels levels before I start to actually balance the mix. But maybe I'm missing something?

[robussc]

This seems like the right approach to me and is very similar to how I go about it. It's all about making sure you've got some headroom when everything comes together, so hitting at -10db is about right. I'll use an instrument volume to get the patch into the rough ball park (unless it's a VST) and then use the input gain to fine tune it as necessary.


Unless you're mixing in live audio you shouldn't have too much worry about noise (and even then, it's nothing compared to the old days with 100% analog gear!)

[selig]

Forget gain staging for a second and lets just talk about levels. This isn’t about noise or “healthy levels” since you have such a WIDE range of legal levels in a floating point system. I prefer to think more about simplifying workflow and removing as many potential surprises as possible.

For more details:

Your approach is not going to do much since the channel meters have TWO issues to deal with. #1, they are not peak meters, so even a fairly low setting can still be clipping the output. For another, if you’re using the default settings, then -10dB on the channel meters is actually 12dB LOWER than what the meters read. If you want to see the actual level, go to the Rack View and select the Big Meter tab from the Hardware Interface section (above the Master Section). Take that VU OFFSET knob and crank it all the way up. Now your metering will read accurately BUT it will still be RMS/averaging so will be a lower value than peak.

Why use peak levels? Because in digital systems, headroom is finite and absolute. So at all times you need to be aware of your highest (peak) level because that’s going to indicate how much, if any, headroom is left in the system.

So why not adjust the level as you suggest, at the input to the mixer (Input Gain, I’m assuming?)? Because what if you insert an effect between the instrument and the mixer? Now your levels are not the same as if you inserted it IN the channel (insert section). IF you were to set the level at the SOURCE (highly recommended), you could move that device to ANY position in the signal path and it would behave the same. Also, what if you create a parallel channel? You’ve now made extra work because any parallel channel will ALSO need it’s gain to be adjusted to match the main channel (this would not be the case if you set gain at the source).

So what do I do and why? I started with digital recording back in the mid 1980s. Every single digital audio system I’ve EVER worked with from that day up to this one has peak meters. For 16 bit digital recording we tried to get all levels as close to clipping as possible without going over, about 2dB below clipping was common.
When 24 bit recording was introduced and we had less to worry about, we no longer aimed to “kiss the reds” as it were. Based on comments by Paul Frindle (SSL, Sony Oxford console designer), I dropped my target levels by 10dB to leave headroom and no longer worry about clipping (yay). So now my peak level for all recordings is around -12dBFS. This is perfect, because in Reason the recording meters (which are peak, naturally) are calibrated to show YELLOW from -12dBFS and above, making it easy - just avoid the yellow! But what about synths? Turns out, at least at one point when I built patches for the Factory Sound Bank, we used -12dBFS peaks as our target levels. Great, everyone is on the same page now.

What else is important with regards to levels? Another key workflow concept is to KEEP all levels at the same peak value in the channel signal path. If you add EQ and it raises the level, you need to lower it. Why? A few reasons. One is to allow easy A/B comparisons and to not be fooled by settings that basically just add overall gain (remember, louder ‘sounds better’ even if nothing else is changed). This also allows you to bypass a device completely and not have a level jump. This also allows you to move quickly when setting compressors or amp simulators because the basic level coming into the device is ALWAYS the same!

Finally, leaving 12dB headroom (or more if you have high track counts) has other benefits besides consistency. For one, it means you won’t be chasing the faders as you mix like with many of the so-called “gain staging” videos out there (even from the Propellerheads themselves). If you keep pulling down the master fader as you add more instruments to your mix, you’ve experienced what I’m talking about.

With my “consistent peak levels” approach (I purposely don’t call it gain staging because that term has more meaning and history in the analog domain), you won’t likely ever clip an audio recording again, you also won’t likely clip your mix bus, you can A/B or bypass any processor without totally breaking the mix, you can quickly add gain-dependant processors without major surprises (you can even build your own presets for things like compressors because you already know what levels to expect), etc.

[motuscott]

Selig, as usual (always?) is King of Audio Recording Knowledge.
Selig, your payments are late this month...

[robussc]

Always find it helpful to learn the DAW is not fit for purpose…

[Toki]

Forget gain staging for a second and lets just talk about levels. This isn’t about noise or “healthy levels” since you have such a WIDE range of legal levels in a floating point system. I prefer to think more about simplifying workflow and removing as many potential surprises as possible.

For more details:

Your approach is not going to do much since the channel meters have TWO issues to deal with. #1, they are not peak meters, so even a fairly low setting can still be clipping the output. For another, if you’re using the default settings, then -10dB on the channel meters is actually 12dB LOWER than what the meters read. If you want to see the actual level, go to the Rack View and select the Big Meter tab from the Hardware Interface section (above the Master Section). Take that VU OFFSET knob and crank it all the way up. Now your metering will read accurately BUT it will still be RMS/averaging so will be a lower value than peak.

Why use peak levels? Because in digital systems, headroom is finite and absolute. So at all times you need to be aware of your highest (peak) level because that’s going to indicate how much, if any, headroom is left in the system.

So why not adjust the level as you suggest, at the input to the mixer (Input Gain, I’m assuming?)? Because what if you insert an effect between the instrument and the mixer? Now your levels are not the same as if you inserted it IN the channel (insert section). IF you were to set the level at the SOURCE (highly recommended), you could move that device to ANY position in the signal path and it would behave the same. Also, what if you create a parallel channel? You’ve now made extra work because any parallel channel will ALSO need it’s gain to be adjusted to match the main channel (this would not be the case if you set gain at the source).

So what do I do and why? I started with digital recording back in the mid 1980s. Every single digital audio system I’ve EVER worked with from that day up to this one has peak meters. For 16 bit digital recording we tried to get all levels as close to clipping as possible without going over, about 2dB below clipping was common.
When 24 bit recording was introduced and we had less to worry about, we no longer aimed to “kiss the reds” as it were. Based on comments by Paul Frindle (SSL, Sony Oxford console designer), I dropped my target levels by 10dB to leave headroom and no longer worry about clipping (yay). So now my peak level for all recordings is around -12dBFS. This is perfect, because in Reason the recording meters (which are peak, naturally) are calibrated to show YELLOW from -12dBFS and above, making it easy - just avoid the yellow! But what about synths? Turns out, at least at one point when I built patches for the Factory Sound Bank, we used -12dBFS peaks as our target levels. Great, everyone is on the same page now.

What else is important with regards to levels? Another key workflow concept is to KEEP all levels at the same peak value in the channel signal path. If you add EQ and it raises the level, you need to lower it. Why? A few reasons. One is to allow easy A/B comparisons and to not be fooled by settings that basically just add overall gain (remember, louder ‘sounds better’ even if nothing else is changed). This also allows you to bypass a device completely and not have a level jump. This also allows you to move quickly when setting compressors or amp simulators because the basic level coming into the device is ALWAYS the same!

Finally, leaving 12dB headroom (or more if you have high track counts) has other benefits besides consistency. For one, it means you won’t be chasing the faders as you mix like with many of the so-called “gain staging” videos out there (even from the Propellerheads themselves). If you keep pulling down the master fader as you add more instruments to your mix, you’ve experienced what I’m talking about.

With my “consistent peak levels” approach (I purposely don’t call it gain staging because that term has more meaning and history in the analog domain), you won’t likely ever clip an audio recording again, you also won’t likely clip your mix bus, you can A/B or bypass any processor without totally breaking the mix, you can quickly add gain-dependant processors without major surprises (you can even build your own presets for things like compressors because you already know what levels to expect), etc.

Firstly, thanks for the detailed response and link. I've watched and read through a few times now and so I guess what I've been doing is basically sending some unknown level into the mixer, adjusting that level with the channel input gain and using the channel meter (which isn't accurate) to try to get to -12db, or -10 in my case. Better to keep an eye on Big Meter for accurate readings.

If I understand you correctly, Giles, you're saying the best way to control and maintain consistent levels is to use Selig Gain/Gain Tool between each instrument (+FX) and mixer channel, as well as between the Master Section and Audio I/O. Then use the instruments volume knob to ensure the level into the gain tool is -12db and that it also leaves the gain tool at -12db to the mixer channel, and between -3 to -6 on the Master Section. I guess we could go a step further and put a gain tool between instrument and FX as well as FX and mixer channel if we were so inclined?

So the input gain on the channel strip probably isn't used unless we have actual hardware or live instruments coming into Reason?

[selig]

Firstly, thanks for the detailed response and link. I've watched and read through a few times now and so I guess what I've been doing is basically sending some unknown level into the mixer, adjusting that level with the channel input gain and using the channel meter (which isn't accurate) to try to get to -12db, or -10 in my case. Better to keep an eye on Big Meter for accurate readings.

If I understand you correctly, Giles, you're saying the best way to control and maintain consistent levels is to use Selig Gain/Gain Tool between each instrument (+FX) and mixer channel, as well as between the Master Section and Audio I/O. Then use the instruments volume knob to ensure the level into the gain tool is -12db and that it also leaves the gain tool at -12db to the mixer channel, and between -3 to -6 on the Master Section. I guess we could go a step further and put a gain tool between instrument and FX as well as FX and mixer channel if we were so inclined?

So the input gain on the channel strip probably isn't used unless we have actual hardware or live instruments coming into Reason?

Actually no, I'm not saying to EVER send an unknown level anywhere. You can boil down what I'm saying to mean the opposite - ALWAYS know the level you are sending anywhere. And one of the simplest ways to do that is to always use the same level, because then you will always know what the level is!

But specifically, Selig Gain was partly created for this lack of metering options in Reason anywhere but the master meter and the big meter (I always use VU + Peak because it shows crest factor, which is a good indication of loudness IMO).
The thing is, you don't even have to leave the Selig Gain in place, you can just use it to "spot check" places in the signal path. Then Shift-drag it to a new place and check the levels there, and repeat!
You don't have to be super careful with exact levels either, unless you are doing a critical A/B comparison. Just get within a few dB of the target at the loudest part and you're good. From there, pay attention to your habits and make gradual adjustments. For example, If you use -12dBFS for your peak reference level but still often clip the mix (could be because you have a high track count), try lowering your ref level from -12dBFS to -15dBFS or lower!

And to be clear, this reference level is for individual tracks such as kick snare, bass, lead vocal, etc. For bus levels I often see anywhere from peak around -9dBF up to -6dBFS (for heavy drum busses, as one example). Then for the entire mix, I'm hitting from -3dBFS to -1dBFS and I try not to let the mix peak level go above -3 or -2dBFS on 'raw' mixes just to leave a little safety margin for random crap!

Screen Shot 2024-08-24 at 1.59.04 PM.png

[Toki]

Firstly, thanks for the detailed response and link. I've watched and read through a few times now and so I guess what I've been doing is basically sending some unknown level into the mixer, adjusting that level with the channel input gain and using the channel meter (which isn't accurate) to try to get to -12db, or -10 in my case. Better to keep an eye on Big Meter for accurate readings.

If I understand you correctly, Giles, you're saying the best way to control and maintain consistent levels is to use Selig Gain/Gain Tool between each instrument (+FX) and mixer channel, as well as between the Master Section and Audio I/O. Then use the instruments volume knob to ensure the level into the gain tool is -12db and that it also leaves the gain tool at -12db to the mixer channel, and between -3 to -6 on the Master Section. I guess we could go a step further and put a gain tool between instrument and FX as well as FX and mixer channel if we were so inclined?

So the input gain on the channel strip probably isn't used unless we have actual hardware or live instruments coming into Reason?

Actually no, I'm not saying to EVER send an unknown level anywhere. You can boil down what I'm saying to mean the opposite - ALWAYS know the level you are sending anywhere. And one of the simplest ways to do that is to always use the same level, because then you will always know what the level is!

But specifically, Selig Gain was partly created for this lack of metering options in Reason anywhere but the master meter and the big meter (I always use VU + Peak because it shows crest factor, which is a good indication of loudness IMO).
The thing is, you don't even have to leave the Selig Gain in place, you can just use it to "spot check" places in the signal path. Then Shift-drag it to a new place and check the levels there, and repeat!
You don't have to be super careful with exact levels either, unless you are doing a critical A/B comparison. Just get within a few dB of the target at the loudest part and you're good. From there, pay attention to your habits and make gradual adjustments. For example, If you use -12dBFS for your peak reference level but still often clip the mix (could be because you have a high track count), try lowering your ref level from -12dBFS to -15dBFS or lower!

And to be clear, this reference level is for individual tracks such as kick snare, bass, lead vocal, etc. For bus levels I often see anywhere from peak around -9dBF up to -6dBFS (for heavy drum busses, as one example). Then for the entire mix, I'm hitting from -3dBFS to -1dBFS and I try not to let the mix peak level go above -3 or -2dBFS on 'raw' mixes just to leave a little safety margin for random crap!


Screen Shot 2024-08-24 at 1.59.04 PM.png

Sorry, maybe that was unclear. What I meant was that it appears that this is what I’ve been doing all along. Sending unknown levels into the mixer and trying to adjust them with meters that don’t give an accurate reading i.e. the channel meters.

Your advice was clear, check the levels and be clear what is going into the mixer from each device.

Thanks for confirming about bus and mix levels, too, I’ll definitely take this into consideration. It sounds like a much more solid approach and now I’ve had a chance to modify a track with this in mind and take a look at the result in the Big Meter, the difference is clear.

Thanks again for taking the time to clear that up.

[selig]

Sorry, maybe that was unclear. What I meant was that it appears that this is what I’ve been doing all along. Sending unknown levels into the mixer and trying to adjust them with meters that don’t give an accurate reading i.e. the channel meters.

Your advice was clear, check the levels and be clear what is going into the mixer from each device.

Thanks for confirming about bus and mix levels, too, I’ll definitely take this into consideration. It sounds like a much more solid approach and now I’ve had a chance to modify a track with this in mind and take a look at the result in the Big Meter, the difference is clear.

Thanks again for taking the time to clear that up.

These values are what works for me, not some absolute/accepted "standard" in the industry. And in fact the -12dBFS value is the only one I actually choose/set myself - the rest work themselves out if you don't add or subtract gain at any stage, which is a part of my approach.
Speaking of which, here is an illustration for THAT concept too, although if the amount of change is only a decibel or so I may not bother correcting it at every single stage.

Screen Shot 2024-08-24 at 2.33.56 PM.png

[tanni]

I would like to ask something in this context. In the screenshot from the BigMeter I have set VU and Peak. In this case, am I correct that the average level is -7 db and the peak is -4 db based on full scale? How do I recognize the CREST factor here?

[jam-s]

You're right the Crest factor is just the difference of peak and average: -7 dB to -4 dB --> 3dB

[selig]

I would like to ask something in this context. In the screenshot from the BigMeter I have set VU and Peak. In this case, am I correct that the average level is -7 db and the peak is -4 db based on full scale? How do I recognize the CREST factor here?

This is another confusing aspect of Reason’s metering, and no that is not correct (through no fault of yours) even though you are reading the meters “correctly”. You have to account for the VU offset, which is an arbitrary offset you can adjust yourself but that defaults to a 12dB offset. This is done because (I’m guessing) in the analog world the VU meter shows “0dB” at a point BELOW the clipping point, and in the digital world “0dB” IS the clipping point. However, most ‘real’ VU meters represent a level as much as 20dB or more below the clipping point, and most DAWs hardware is callibrated these days to a level of -18dBFS = 0dBVU - with me so far?

So the “proper” way to read the meters is to either remove the VU offset yourself (with the knob next to the Big Meters), or just ignore the offset scale and use the same scale for both meters. Remember, changing VU Offset has NO effect on the meters or the level, it’s just a visual offset of the numbers on the graphics.

OK, now that we have THAT out of the way, the crest factor in your example is based on a peak level of -4dB and an average level of -19dB or a crest factor of 15dB (which is fairly normal for individual tracks but low for a mastered mix).
Crest factor actually comes quite close to being the same as LUFs and in most cases I don’t even check LUFS for this reason (unless someone requested a specific LUFs which would be rare in audio mixing but can happen in broadcast world more frequently).

The beauty of using the big meter for crest factor readings comes when you have a limiter on your mix (when mastering) so that the highest peaks are right at or slightly below 0dBFS, in which case the average level on the meters IS your crest factor - easy peasy!

[ReasonUser]

Selig, THANK YOU for this valuable information! I found this gain staging video a few years ago and it was such a huge help for me.

[tanni]

I would like to ask something in this context. In the screenshot from the BigMeter I have set VU and Peak. In this case, am I correct that the average level is -7 db and the peak is -4 db based on full scale? How do I recognize the CREST factor here?

This is another confusing aspect of Reason’s metering, and no that is not correct (through no fault of yours) even though you are reading the meters “correctly”. You have to account for the VU offset, which is an arbitrary offset you can adjust yourself but that defaults to a 12dB offset. This is done because (I’m guessing) in the analog world the VU meter shows “0dB” at a point BELOW the clipping point, and in the digital world “0dB” IS the clipping point. However, most ‘real’ VU meters represent a level as much as 20dB or more below the clipping point, and most DAWs hardware is callibrated these days to a level of -18dBFS = 0dBVU - with me so far?

So the “proper” way to read the meters is to either remove the VU offset yourself (with the knob next to the Big Meters), or just ignore the offset scale and use the same scale for both meters. Remember, changing VU Offset has NO effect on the meters or the level, it’s just a visual offset of the numbers on the graphics.

OK, now that we have THAT out of the way, the crest factor in your example is based on a peak level of -4dB and an average level of -19dB or a crest factor of 15dB (which is fairly normal for individual tracks but low for a mastered mix).
Crest factor actually comes quite close to being the same as LUFs and in most cases I don’t even check LUFS for this reason (unless someone requested a specific LUFs which would be rare in audio mixing but can happen in broadcast world more frequently).

The beauty of using the big meter for crest factor readings comes when you have a limiter on your mix (when mastering) so that the highest peaks are right at or slightly below 0dBFS, in which case the average level on the meters IS your crest factor - easy peasy!

ah, thank you very much for the clarification. Yes, 3 db CREST factor to -4 db full scale couldn't be possible here either, my track would have to sound completely different in this case. I dont have any compression or limiters yet. Average level at -19 dB is more consistent. The way you described it fits the context correctly, I think. Now I can understand that and see the Bigmeter with completely different eyes .

[jam-s]

Yup, Selig is right, of course. I forgot to take the VU offset into account.

[RobC]

I stopped reading at setting loudness by SSL meters. That's a no-go. Meters just give a peak/average loudness information. Everything goes by ear, first.

If you want an equal loudness for everything, first, have a reference sound. Do an A/B leveling with every sound you have. Always listen separately, with a bit of silent time gap. Set your levels with the fader. Done. Simple as that. (Assuming that your sounds/tracks/recordings/etc. are processed and balanced).

The reference sound can be anything (in my case at least, since I have death-accurate critical hearing). Well, perhaps it should be something that's a bit more varied (think pure sine wave vs. filtered noise). Basically, you can even use a recorded, normalized fart as a reference. xD Maybe add 6 dB headroom or so to it.

[selig]

I stopped reading at setting loudness by SSL meters. That's a no-go. Meters just give a peak/average loudness information. Everything goes by ear, first.

If you want an equal loudness for everything, first, have a reference sound. Do an A/B leveling with every sound you have. Always listen separately, with a bit of silent time gap. Set your levels with the fader. Done. Simple as that. (Assuming that your sounds/tracks/recordings/etc. are processed and balanced).

The reference sound can be anything (in my case at least, since I have death-accurate critical hearing). Well, perhaps it should be something that's a bit more varied (think pure sine wave vs. filtered noise). Basically, you can even use a recorded, normalized fart as a reference. xD Maybe add 6 dB headroom or so to it.

There may be some confusion here…we are talking about levels (which are measured by meters), not loudness.

For the record, my post is the ONLY post that mentions “loudness”, and you’re taking that mention out of context…
No one is suggesting “equal loudness for everything” anywhere in this thread.
I’ve always set my levels at the SOURCE, and set my mix BALANCES with the faders.
Hope that clarifies things?

[RobC]

I stopped reading at setting loudness by SSL meters. That's a no-go. Meters just give a peak/average loudness information. Everything goes by ear, first.

If you want an equal loudness for everything, first, have a reference sound. Do an A/B leveling with every sound you have. Always listen separately, with a bit of silent time gap. Set your levels with the fader. Done. Simple as that. (Assuming that your sounds/tracks/recordings/etc. are processed and balanced).

The reference sound can be anything (in my case at least, since I have death-accurate critical hearing). Well, perhaps it should be something that's a bit more varied (think pure sine wave vs. filtered noise). Basically, you can even use a recorded, normalized fart as a reference. xD Maybe add 6 dB headroom or so to it.

There may be some confusion here…we are talking about levels (which are measured by meters), not loudness.

For the record, my post is the ONLY post that mentions “loudness”, and you’re taking that mention out of context…
No one is suggesting “equal loudness for everything” anywhere in this thread.
I’ve always set my levels at the SOURCE, and set my mix BALANCES with the faders.
Hope that clarifies things?

Sure, levels and loudness are both important to pay attention to, but it wasn't clear if we talk about normalizing, or setting volume by ear.

Sorry, I just read the OP, partly... : S

Personally, I'm like, process/balance and normalize samples/recordings/tracks/etc. and then set levels in context with the mix (as well as further processing, if needed).
So the way I think is, sounds on their own; and then sounds in context with the mix.

But I'll read the whole thread now, cause I probably got things wrong, I guess.

[RobC]

I stopped reading at setting loudness by SSL meters. That's a no-go. Meters just give a peak/average loudness information. Everything goes by ear, first.

If you want an equal loudness for everything, first, have a reference sound. Do an A/B leveling with every sound you have. Always listen separately, with a bit of silent time gap. Set your levels with the fader. Done. Simple as that. (Assuming that your sounds/tracks/recordings/etc. are processed and balanced).

The reference sound can be anything (in my case at least, since I have death-accurate critical hearing). Well, perhaps it should be something that's a bit more varied (think pure sine wave vs. filtered noise). Basically, you can even use a recorded, normalized fart as a reference. xD Maybe add 6 dB headroom or so to it.

There may be some confusion here…we are talking about levels (which are measured by meters), not loudness.

For the record, my post is the ONLY post that mentions “loudness”, and you’re taking that mention out of context…
No one is suggesting “equal loudness for everything” anywhere in this thread.
I’ve always set my levels at the SOURCE, and set my mix BALANCES with the faders.
Hope that clarifies things?

So, if I got it right, this seems to be rather a "worry" about peaks and clipping, versus noise floor.

I mean... assuming a professional recording and mixing, there indeed would be a ~6-12 dB headroom (versus a mix, aka. a finished song mix), which in the digital realm doesn't matter much to begin with (as back a bunch of years, the Props said, too, "don't sweat it").
See, it doesn't matter if I record, or synthesize, I always do processing if/as needed. A sound (literally a one shot sample; a loop; a recording; anything) would end up processed and normalized (unless it's live). During mixing, I can confidentally toss in everything into the main mixer, maybe pull faders back if needed (though the main fader is enough to pull back a bit, too); and get to the mixing work.

Personally, I don't get what's so complicated about it to people. It's super simple. Just two rules: stay below clipping, and above the noise floor. (The latter tops being trurly important when recording anything from the analog realm / "real world").

[selig]

…Personally, I don't get what's so complicated about it to people. It's super simple. Just two rules: stay below clipping, and above the noise floor. (The latter tops being trurly important when recording anything from the analog realm / "real world").

I add a few more “rules” to that because of how I like to work, which is consistent peak levels for all sources and don’t add/subtract to this with any processing between the source and the fader. Read my initial post on WHY I do this if you’re interested.

[RobC]

…Personally, I don't get what's so complicated about it to people. It's super simple. Just two rules: stay below clipping, and above the noise floor. (The latter tops being trurly important when recording anything from the analog realm / "real world").

I add a few more “rules” to that because of how I like to work, which is consistent peak levels for all sources and don’t add/subtract to this with any processing between the source and the fader. Read my initial post on WHY I do this if you’re interested.

I read it again, and I think I work in a similar way then. Maybe I'm not that fixed on what the peak level has to be, because I preferred to be above the noise floor as much as possible. I pretty much exclusively use mixers for leveling, and nothing else. The rest happens in the rack for example. Although my mixing strategy changes, in the past, with the normalized approach, the loudest sound's fader always was at 100 out of 127. Since Reason's mixers don't clip, I just pulled the master fader back, and I was good to go.
The peaks were always constant, cause I always used a limiter/clipper before the sound went into the mixer.
We used to call it "pre-mastering" each track individually.
If I put a pack up for people for remixing, they always said, that it was the most organized and easiest to work with and easiest to mix.
Personally, I believe if tracks are rendered, and normalized, and I do the mixing inside Reason, then it's enough to pull the master fader back, cause it's clipping safe.
That said, nowadays, the way I mix, will change drastically. Because with my IEM system amplified to the max, I will just have a reference sound set to a comfortable level, which probably will have like 24 or more headroom to begin with, when normalized, so there definitely won't be any surprises. (Of course, analog recording will be a whole other story.)

But yeah, if I get it right, this gain staging is pretty much a waste of time. You can say the same for levelling sounds to a reference, so they are heardas equally loud : ) but that has its own purposes.

[selig]

I add a few more “rules” to that because of how I like to work, which is consistent peak levels for all sources and don’t add/subtract to this with any processing between the source and the fader. Read my initial post on WHY I do this if you’re interested.

I read it again, and I think I work in a similar way then. Maybe I'm not that fixed on what the peak level has to be, because I preferred to be above the noise floor as much as possible. I pretty much exclusively use mixers for leveling, and nothing else. The rest happens in the rack for example. Although my mixing strategy changes, in the past, with the normalized approach, the loudest sound's fader always was at 100 out of 127. Since Reason's mixers don't clip, I just pulled the master fader back, and I was good to go.
The peaks were always constant, cause I always used a limiter/clipper before the sound went into the mixer.
We used to call it "pre-mastering" each track individually.
If I put a pack up for people for remixing, they always said, that it was the most organized and easiest to work with and easiest to mix.
Personally, I believe if tracks are rendered, and normalized, and I do the mixing inside Reason, then it's enough to pull the master fader back, cause it's clipping safe.
That said, nowadays, the way I mix, will change drastically. Because with my IEM system amplified to the max, I will just have a reference sound set to a comfortable level, which probably will have like 24 or more headroom to begin with, when normalized, so there definitely won't be any surprises. (Of course, analog recording will be a whole other story.)

But yeah, if I get it right, this gain staging is pretty much a waste of time. You can say the same for levelling sounds to a reference, so they are heardas equally loud : ) but that has its own purposes.

I pay attention to peak levels for the same reasons you do not, meaning I pay attention to peak levels because I want to avoid noise and avoid clipping. I also pay attention to headroom, and you can’t know your headroom without knowing peak levels. If you’re lowering your master fader you’re likely hitting your mix bus too hot - I prefer hitting it at a level that when I bypass everything (including master fader) my levels are perfect for mastering. Consistent track levels also means consistent bus levels, so I’m able to quickly apply settings throughout the mix because I already know what level things are at - make sense? Otherwise, I’d be chasing the master fader too, I guess. I keep levels consistent as I add processing, again for quickness but also so that if I bypass something the levels don’t jump. Just trying to keep things as simple as possible, and consistency helps because it’s one less thing I need to figure out for every track etc.

I also don’t advocate leveling sounds to a reference, not sure why that keeps coming up in this discussion?

[motuscott]

I once knew a guy/gal who would just ride that fader into oblivion.
Some of you analogue types know what I'm sain.
If it ain't red, you're not dead.
How long til Mods attack me cos of non political rebeefs, guessing 20 nanules,
Any body up for that?

[Theo.M]

In PT because the clip gain is so good, I actually use that a lot and automate that rather than the fader. Once you use it you would understand why. All you need to do is select a part of audio and point are auto created, you can keep things relative as well, and I even use it for manual de essing cause it's so easy. ZERO clip chopping required.

I also like to use the K12 system mostly, and use a K12 meter:
https://www.meterplugs.com/kmeter

Basically all that's really doing is setting minus 12 dbfs as 0, so if something is hitting 0 on the K meter, it will be peaking at -12 on the typical DAW channel meter. It's a simple system Bob Katz devised to help combat the loudness war. Gives good headroom for any type of music. My aim is strictly online, no physical media, so I am going for -14 LUFS and -16 for Apple.

I find the K system works for me, I always have a gentle ssl on the master buss (Cytomic) even as I am composing, from the get go, and for the actual gain control where I am not just using clip gain, or non audio track channels, I use VUMT Deluxe:

https://klanghelm.com/contents/products/VUMT

In Reason I used Selig Gain a lot, but I am exclusively using Reason as an AAX plugin now.

[RobC]

I read it again, and I think I work in a similar way then. Maybe I'm not that fixed on what the peak level has to be, because I preferred to be above the noise floor as much as possible. I pretty much exclusively use mixers for leveling, and nothing else. The rest happens in the rack for example. Although my mixing strategy changes, in the past, with the normalized approach, the loudest sound's fader always was at 100 out of 127. Since Reason's mixers don't clip, I just pulled the master fader back, and I was good to go.
The peaks were always constant, cause I always used a limiter/clipper before the sound went into the mixer.
We used to call it "pre-mastering" each track individually.
If I put a pack up for people for remixing, they always said, that it was the most organized and easiest to work with and easiest to mix.
Personally, I believe if tracks are rendered, and normalized, and I do the mixing inside Reason, then it's enough to pull the master fader back, cause it's clipping safe.
That said, nowadays, the way I mix, will change drastically. Because with my IEM system amplified to the max, I will just have a reference sound set to a comfortable level, which probably will have like 24 or more headroom to begin with, when normalized, so there definitely won't be any surprises. (Of course, analog recording will be a whole other story.)

But yeah, if I get it right, this gain staging is pretty much a waste of time. You can say the same for levelling sounds to a reference, so they are heardas equally loud : ) but that has its own purposes.

I pay attention to peak levels for the same reasons you do not, meaning I pay attention to peak levels because I want to avoid noise and avoid clipping. I also pay attention to headroom, and you can’t know your headroom without knowing peak levels. If you’re lowering your master fader you’re likely hitting your mix bus too hot - I prefer hitting it at a level that when I bypass everything (including master fader) my levels are perfect for mastering. Consistent track levels also means consistent bus levels, so I’m able to quickly apply settings throughout the mix because I already know what level things are at - make sense? Otherwise, I’d be chasing the master fader too, I guess. I keep levels consistent as I add processing, again for quickness but also so that if I bypass something the levels don’t jump. Just trying to keep things as simple as possible, and consistency helps because it’s one less thing I need to figure out for every track etc.

I also don’t advocate leveling sounds to a reference, not sure why that keeps coming up in this discussion?

There may be a misunderstanding. I likewise watch it to record/render with just enough (plausible) "headroom", avoiding clipping, yet not drowning in the noisefloor either.
If I normalize such recorded/rendered audio, then the peak will be at 0 dB.

As far as I know, Reason works internally at 32 bit depth, with 64 bit summing in the main mixer. So, while working inside Reason, there's no need to worry. Sure, we can play it safe and practice keeping safe volume levels, but even if a mix channel is overblown, it won't ever clip.
If we want to set the master fader to 0 dB, we can just select every fader, and pull them back until it won't clip and there's enough headroom.

On mixers that clip, yes, I likewise would leave a safe 12 dB headroom (once I do know my peaks). However, I wouldn't render anything at -12 dB, because then it's already that much buried into the noise floor (though we can get away with it even at 16 bits).

That said, sometimes we want to go back and change something on a sound, which may be drastic enough, that we have to re-level it.
I feel that if we add a limiter, or render something for a mix channel, then we limit ourselves.

When I said that I experiment with a fully amplified system, I didn't mean that I'd ever render anything that tops peaks at, say ~ -30 dB. Be it a sound, a mix, or the master. When done, I'd of course compensate the volume.

As for leveling to a reference, I just mentioned that as an experimental thing. That's another, complicated topic.