Gain Staging - Reapers Meter vs FreeG vs SPAN - I'm lost

[Tesgin]

Originally Posted by trevlyns

Just as a side, an interesting post by our Kenny on the Groove 3 facebook site this week:

Kenny’s Tip of the Day - Gain Staging in a DAW does NOT Matter!!!
For years we’ve been living with this myth that the rules that applied to analog recording situations must also apply to digital ones. Or Gain Staging inside our computers or DAWs ...

... So to recap. Gain Staging in a DAW does NOT Matter!!! It’s an outdated philosophy that has hung around for far too long. Do away with it and get back to doing what’s important.
Create something!!
You can learn more by watching my tutorials at www.groove3.com
I hope this message finds you well. Kenny Gioia

Great find. That makes perfect sense to me: The important things are recording at the right level (a lot of headroom) and the levels at the master bus (no clipping).

Thanks,
TB

[Neenja]

Originally Posted by Tesgin

Great find. That makes perfect sense to me: The important things are recording at the right level (a lot of headroom) and the levels at the master bus (no clipping).

Thanks,
TB

It matters in the case of the OP and with old plugins that aren't 32 bit. It really should have been titled "gain staging in DAWS doesn't matter now".

[Tesgin]

This is probably a little off topic, but my question arises out of this thread ...

In the master vu settings, I understand the display offset, display gain and red threshold mean; I don't understand what window size refers to. Not intuitive, cuz the measurement is in ms, which doesn't pertain to size at all.

I couldn't find anything in the manual about this.

Could someone explain that to me?

TB

[Judders]

Originally Posted by Tesgin

This is probably a little off topic, but my question arises out of this thread ...

In the master vu settings, I understand the display offset, display gain and red threshold mean; I don't understand what window size refers to. Not intuitive, cuz the measurement is in ms, which doesn't pertain to size at all.

I couldn't find anything in the manual about this.

Could someone explain that to me?

TB

Are you talking about the RMS meter on the master channel? Window size is the time that the average level is taken over.

[Tesgin]

Originally Posted by Judders

Are you talking about the RMS meter on the master channel? Window size is the time that the average level is taken over.

Yes, that is what I was referring to. The master meter settings.

I'm not following you. What does that mean? The time the average level has taken over? I don't understand.

I "think" it means the window of time in ms that is used to calculate the average RMS? So if window size is 500 ms, that means that the RMS reading is calculated continuously based on 500 ms intervals.

Am I close?

Thanks,

TB

[RJHollins]

look up 'specification of RMS meters'

[Judders]

Originally Posted by Tesgin

Yes, that is what I was referring to. The master meter settings.

I'm not following you. What does that mean? The time the average level has taken over? I don't understand.

I "think" it means the window of time in ms that is used to calculate the average RMS? So if window size is 500 ms, that means that the RMS reading is calculated continuously based on 500 ms intervals.

Am I close?

Thanks,

TB

Yes, you got it!

[chas51]

I do not understand this part of Kenny's post:

"So what does this mean? It means that “proper gain staging” is mostly irrelevant in a DAW. Anything you Clip at one stage can be fixed further down the line. You don’t need to check the level on every single channel, buss or plugin to make sure it’s operating at peak efficiency. It is. And if you are clipping somewhere, you can always fix it later."

I do not understand that clipping is ok, and that you can fix it later.
would someone please tell me if they have any thoughts about this?

[karbomusic]

I do not understand that clipping is ok, and that you can fix it later.
would someone please tell me if they have any thoughts about this?

Internally the audio is processed at 64 bit float IIRC, this means that no matter what the track meters say, the audio can't clip as long as that stays inside the box. Thusly if you do what looks like clipping the crap out of it on a track, but reduce it later in the audio chain, and before the D/A converters (aka the master, folder or buss), it will be just fine and not actually clipped.

Additionally, If the glue settings are at least 32 bit float, you can even render/glue a file that is 100 dB above zero, then load the file, reduce it by 100 dB and it won't be damaged/clipped.

[Neenja]

Originally Posted by chas51

I do not understand this part of Kenny's post:

"So what does this mean? It means that “proper gain staging” is mostly irrelevant in a DAW. Anything you Clip at one stage can be fixed further down the line. You don’t need to check the level on every single channel, buss or plugin to make sure it’s operating at peak efficiency. It is. And if you are clipping somewhere, you can always fix it later."

I do not understand that clipping is ok, and that you can fix it later.
would someone please tell me if they have any thoughts about this?

That explanation isn't accurate. In Reaper and modern DAWs, you can push volumes past zero on the meters and they don't clip. You still have to make sure you don't record past zero or render to less than 32bit while exceeding zero.

[chas51]

Originally Posted by karbomusic

Internally the audio is processed at 64 bit float IIRC, this means that no matter what the track meters say, the audio can't clip as long as that stays inside the box. Thusly if you do what looks like clipping the crap out of it on a track, but reduce it later in the audio chain, and before the D/A converters (aka the master, folder or buss), it will be just fine and not actually clipped.

Additionally, If the glue settings are at least 32 bit float, you can even render/glue a file that is 100 dB above zero, then load the file, reduce it by 100 dB and it won't be damaged/clipped.

thank you very much karbomusic.
if you feel like it, would you please say more?
do you mean that in Reaper, there is infinite headroom?
and I can record as hot as I want, and not worry about clipping and digital distortion, even though I recorded hotter than 0 dB?
I'm just not getting it, man.

Originally Posted by Neenja

That explanation isn't accurate. In Reaper and modern DAWs, you can push volumes past zero on the meters and they don't clip. You still have to make sure you don't record past zero or render to less than 32bit while exceeding zero.

thank you Neenja.

[Judders]

Originally Posted by chas51

thank you very much karbomusic.
if you feel like it, would you please say more?
do you mean that in Reaper, there is infinite headroom?
and I can record as hot as I want, and not worry about clipping and digital distortion, even though I recorded hotter than 0 dB?
I'm just not getting it, man.

INSIDE Reaper there is basically infinite headroom (not really, but effectively for any sane usage).

Going INTO or OUT OF Reaper there is not. So no, you cannot record as hot as you want, because that will clip the analogue-to-digital converter in your audio interface. Same as you can't output Reaper's master channel as hot as you want, because it feeds a 24 bit signal to your audio interface's digital-to-analogue converter.

But for most processing duties you perform on audio that is already recorded and in Reaper, you effectively can't clip (with the exception of some analogue emulation plugins that simulate analogue headroom).

[chas51]

ah, thank you Judders.
so lets talk about on the way INTO Reaper.
you're recording yourself singing through a microphone. just one microphone.
you are singing and accompanying yourself on acoustic guitar.
if your peaks on the Track meter do not go higher than -10 or -8 dB, you should be just fine.
does that seem good to you?

[Judders]

Originally Posted by chas51

ah, thank you Judders.
so lets talk about on the way INTO Reaper.
you're recording yourself singing through a microphone. just one microphone.
you are singing and accompanying yourself on acoustic guitar.
if your peaks on the Track meter do not go higher than -10 or -8 dB, you should be just fine.
does that seem good to you?

Yep, stick to that and you're golden

[chas51]

Originally Posted by Judders

Yep, stick to that and you're golden

thank you very much again!!!

[Judders]

Originally Posted by chas51

thank you very much again!!!

You're very welcome!

[ashcat_lt]

OUTSIDE of Reaper, Reaper's meters are meaningless. You should be shooting for levels that keep your analog gear happy. Good gain staging still matters in the analog world.

Most of the time, that still means you need to mess around to get the best signal-to-noise ratio without unwanted distortion. Adjust your gains with that in mind and let it fall where it may in Reaper unless it ends up clipping.


True, with the vast majority of interfaces that we are using, we don't have any other meters than the ones in Reaper. Maybe an LED that turns green for signal and red for clipping. But you still shouldn't be setting your gains based on Reaper's meters.

The truth is that gain staging in any situation should never be done by eye. No meters really matter. You set your gain staging by ear. If you're hearing more noise than you'd like, or you're getting distortion that doesn't sound appropriate, you adjust things until it gets as good as it can.

As long as it's as good as it gets going in, and it never hits 0dbfs, it's as good as it's going to get whether it's peaking at -0.001 (lol, I think this actually lights reaper's clip indicator, though it's technically not) or -60. Like, the dynamic range isn't going to change after that point unless and until you deliberately apply processing to do so.

I will say that I think gain staging your output is important. I've never bothered really calibrating my monitors for SPL, but I do have it set up so that even if the output hits 0dbfs, it's never quite going to hit the limits of the monitor controller or amplifiers. This way if I hear distortion, I know it's in the mix, not coming from my monitors.

[chas51]

thank you.

[buenavista]

Originally Posted by trevlyns

Just as a side, an interesting post by our Kenny on the Groove 3 facebook site this week:

Kenny’s Tip of the Day - Gain Staging in a DAW does NOT Matter!!!
For years we’ve been living with this myth that the rules that applied to analog recording situations must also apply to digital ones. Or Gain Staging inside our computers or DAWs.
And this one has hung around for decades at this point. I’m assuming it’s because older engineers (like me) have come from the analog world and it was always good practice to have proper levels at EVERY stage of the process.
So if I was recording to Tape, I would need to get the perfect recording level to that tape. And that had a very small tolerance. Too hot and it distorted, but too low and it was noisy. Or full of hiss. That also no longer exists. True. We can still distort if we record too hot (clipping) but if we record too low, there’s really NO noise to worry about. So it matters very little to get this part perfect. There’s plenty of wiggle room here. But we’ll get back to this in a bit.
After hitting the Tape Machine, that sound would come back to the console where we would need to adjust the Line Trim on each channel. Making sure we weren’t overloading the console at that point.
Next we had inserts for adding effects. You needed to be careful to get the level TO the effects correctly and they had input levels as well. And you had to come back out of them at the right level again because now you were going back to the console. So each of these levels need to be just right. Consoles also added their own noise which needed to be avoided.
Once we finally left the channel, we then are hitting our busses. Again, we needed to check our levels here to avoid overloading them too. Then there might be Sub-Groups and finally we had the Master 2 Buss or the Master Fader. That too had signal coming to it and also had to be gain staged properly. This was no small task and it became a very important part of the engineer’s job. So much so, that it became almost subconscious. We did it without thinking. Eventually.
Then came DAWs and we brought this process to the computer as well. The DAW also had adjustments at all of the same places. So we continued to do what we considered to be “best practice”. After all. It wasn’t hurting anyone. Use it. Teach it. Who cares?
And that’s true for the most part but it’s also not accurate to the equipment we’re now using. I remember a thread going back over a decade ago where a guy suggested his mixes sounded so much “warmer” or more “analog” when he started Trimming each track down -18dB before he started mixing. That thread went on for 100s and 100s of pages and I believe the consensus was that it worked!!!! Ugh.
Let’s talk truth. Unlike with analog systems. Digital systems don’t care about level. In other words, if the signal is -2dB or -30db, the signal will be identical except for those volume differences. Noise and Distortion do NOT change in the audio. So you can mix at -30dB and boost that final mix by 28dB and you will have the same exact same mix as the -2dB mix. Digital doesn’t care where or when you adjust it. It figures it out and makes the proper adjustments.
So what does this mean? It means that “proper gain staging” is mostly irrelevant in a DAW. Anything you Clip at one stage can be fixed further down the line. You don’t need to check the level on every single channel, buss or plugin to make sure it’s operating at peak efficiency. It is. And if you are clipping somewhere, you can always fix it later.
So this doesn’t mean that there are NO rules. Just that there are a lot less of them.
Rule #1 - Record at conservative levels. There’s no reason to record as hot as possible and many reasons not to. Noise is no longer a factor so the optimal level of most input equipment (your mic preamp and converter) is -18dB. That’s where your front end works best. You can go a bit hotter with peaks in the -10dB to -8dB range but anything more and you’re pushing your preamps and convertors out of their preferred range. And for no reason. So record lower.
Rule #2 - Once you’re in the DAW or the computer - All bets are OFF!! Go crazy. Normalize your levels. Bring up their gain. Boost everything as much as you want. It doesn’t matter anymore as long as you follow rule number three.
Rule #3 - Pay attention to your Output Level. Is that clipping? If it is, bring it down. Use a trimming or volume plugin or just lower the Master Fader. As long as this level is not clipping to your Digital to Analog Convertor (which is your final output stage) you’re GOLD. It’s all good. You’ve done your job.
Exception to the above rules. If you use an emulation plugin where the manufacturer tells you to hit that plugin with a certain level to get the best out of it, do as they say. But if it’s a quality plugin, it should have an input gain on that plugin to achieve that level and a good meter.
So to recap. Gain Staging in a DAW does NOT Matter!!! It’s an outdated philosophy that has hung around for far too long. Do away with it and get back to doing what’s important.
Create something!!
You can learn more by watching my tutorials at www.groove3.com
I hope this message finds you well. Kenny Gioia

When I saw the video that Kenny made explaining this, it was kind of an "aha" moment for me. I've always been very conscientious about recording levels and headroom in my analog gear, and always just applied the same rules to the DAW.

I have one question though. When you adjust a fader on an audio track in Reaper to apply additional volume beyond the originally recorded material, doesn't that "add" digital information? In other words, that is in itself a gain stage isn't it? It may not matter from a workflow perspective, or clipping perspective, but are there any doubters out there about the quality of digital amplification? Or am I missing something about the floating point system when I think that there must be an additive process here?

[Judders]

Originally Posted by buenavista

When I saw the video that Kenny made explaining this, it was kind of an "aha" moment for me. I've always been very conscientious about recording levels and headroom in my analog gear, and always just applied the same rules to the DAW.

I have one question though. When you adjust a fader on an audio track in Reaper to apply additional volume beyond the originally recorded material, doesn't that "add" digital information? In other words, that is in itself a gain stage isn't it? It may not matter from a workflow perspective, or clipping perspective, but are there any doubters out there about the quality of digital amplification? Or am I missing something about the floating point system when I think that there must be an additive process here?

Once you get to 64 bit floating point, as with REAPER, then the errors produced are so insignificant as to be not worth worrying about.

In the old days of fixed integer 16, 24 or 48 bit digital summing, then dithering would often be used to prevent truncation distortion.

[ashcat_lt]

Originally Posted by buenavista

I have one question though. When you adjust a fader on an audio track in Reaper to apply additional volume beyond the originally recorded material, doesn't that "add" digital information?

No. It multiplies.

Edit - didn't actually mean to post that quick!

It's kind of rough math stuff, but you can neither gain not lose resolution via straight multiplication until you hit a limit on one end or another.

Consider the series of integers between 1 and 10. It includes 10 unique members covering a span of 9 units. Multiply each by 2. Now it spans 18 units, but still only has 10 members. Likewise, if you divide each in half, it now spans 4.5 units, but still has 10 members.

BUT if you can't get bigger than 10, the multiply only has 5 unique members and only spans 9 units. And similarly, if you can't have anything smaller than 1, the divide has less unique members and smaller total span.

That's over-simplified, but hopefully helps?

It becomes weirder when you start talking about non-linear processes which can actually create numbers "in between", but we have a lot more resolution in floating point than any analog system anywhere, and unless you're doing really extreme things, you'll never even measure an error you could blame on that.

[JSMastering]

Originally Posted by buenavista

When I saw the video that Kenny made explaining this, it was kind of an "aha" moment for me. I've always been very conscientious about recording levels and headroom in my analog gear, and always just applied the same rules to the DAW.

I have one question though. When you adjust a fader on an audio track in Reaper to apply additional volume beyond the originally recorded material, doesn't that "add" digital information? In other words, that is in itself a gain stage isn't it? It may not matter from a workflow perspective, or clipping perspective, but are there any doubters out there about the quality of digital amplification? Or am I missing something about the floating point system when I think that there must be an additive process here?

Without talking about math, and putting it in analog term...

Reaper doesn't have a noise floor.
It also doesn't have voltage rails.

You're not amplifying noise when you turn stuff up, and you're not going to clip it until you hit the DAC.

Analog-modeling plugins may have either or both of those things and/or a preferred operating level depending on how they were designed.

[Philbo King]

Originally Posted by Judders

The only exception is analogue emulation plugins that are level-dependent. But even then they all have input knobs, so you just twist that till you get the amount of distortion you want.

I think some people use DAW's like a classic studio simulation game, rather than a tool to make music.

Y'know, like those nerds who got Microsoft Flight Simulator just to fly real passenger routes to regulation standards, and it never occurred to them to try and take out Sears Tower.

Hah - That takes me back. I wrote a 747 flight simulator in MBASIC once, long ago, in another life. Sometime in the 80s, back when I was teaching college...
It even had a mockup of the radio systems. You had to enter VOR frequencies to navigate , use the glide slope indicators to land, all that stuff.

Just an old guy reminiscing...

[Judders]

Originally Posted by Philbo King

Hah - That takes me back. I wrote a 747 flight simulator in MBASIC once, long ago, in another life. Sometime in the 80s, back when I was teaching college...
It even had a mockup of the radio systems. You had to enter VOR frequencies to navigate , use the glide slope indicators to land, all that stuff.

Just an old guy reminiscing...

Übernerd!

[Not_Here]

Microsoft Flight Sim on an IBM PC XT...
That was the stuff right there! Used to try to fly through impossible weather. Oh the graphics...

[Philbo King]

Originally Posted by Judders

Übernerd!

That's me!!!

[buenavista]

Originally Posted by JSMastering

Without talking about math, and putting it in analog term...

Reaper doesn't have a noise floor.
It also doesn't have voltage rails.

You're not amplifying noise when you turn stuff up, and you're not going to clip it until you hit the DAC.

Analog-modeling plugins may have either or both of those things and/or a preferred operating level depending on how they were designed.

If there is no noise floor, and the digital amplification is basically unnoticeable, then what is to keep someone from making an ADC that can handle the impedance of microphone level. In other words, aside from the character that a preamp may impart, why do we need preamps at all? Compress on the way in? There are near zero latency compressors to do that job if you want. Want character? . . . there are flight simulators that can do that. :-)

[JSMastering]

Originally Posted by buenavista

If there is no noise floor, and the digital amplification is basically unnoticeable, then what is to keep someone from making an ADC that can handle the impedance of microphone level. In other words, aside from the character that a preamp may impart, why do we need preamps at all? Compress on the way in? There are near zero latency compressors to do that job if you want. Want character? . . . there are flight simulators that can do that. :-)

Reaper doesn't have a noise floor because it uses floating point math.

ADCs do because they're relating analog voltage levels to fixed point numbers. Also because the analog parts of the ADC have noise floors.

In SOFTWARE, most of the time, with modern stuff, the level doesn't matter. When you get to ADCs and DACs, it does.

[buenavista]

Could you extrapolate on that? I still don't understand why you couldn't design an ADC to work on microphone levels. What's the original purpose of a preamp? Tape hiss, vinyl noise? . . . Couldn't an ADC theoretically preserve the signal to noise ratio from a microphone as good or better than a preamp?

EDIT: meant to say: preserve the microphone signal without additional analog amplification.

[karbomusic]

Originally Posted by buenavista

Could you extrapolate on that? I still don't understand why you couldn't design an ADC to work on microphone levels.

I get what you are thinking but turning up a signal that is close to the noise floor, in this case a microphone signal, still brings the noise up with it whether you do that before or after the ADC because the noise got recorded too. That said, an ADC that can work on microphone levels, is still electronically an ADC with a preamp circuit in front; all we are doing is changing the description, not the function.

[JSMastering]

Originally Posted by buenavista

Could you extrapolate on that? I still don't understand why you couldn't design an ADC to work on microphone levels. What's the original purpose of a preamp? Tape hiss, vinyl noise? . . . Couldn't an ADC theoretically preserve the signal to noise ratio from a microphone as good or better than a preamp?

EDIT: meant to say: preserve the microphone signal without additional analog amplification.

You could. No one does it because then you wouldn't be able to use it for line level signals.

The whole world settled on specific voltage references for line level, and that's what ADCs work with.

[buenavista]

Originally Posted by karbomusic

I get what you are thinking but turning up a signal that is close to the noise floor, in this case a microphone signal, still brings the noise up with it whether you do that before or after the ADC because the noise got recorded too.

Well, I don't think there is anything special about a signal that is close to the noise floor. That noise floor will always be brought up in an amplification stage in tandem with the intended signal, no matter how close they are to each other.

[/QUOTE]That said, an ADC that can work on microphone levels, is still electronically an ADC with a preamp circuit in front; all we are doing is changing the description, not the function. [/QUOTE]

Changing the description would change the function because the function would be to convert an audio signal without the need for a preamp. And furthermore, convert an audio signal without the added noise that a preamp makes when it boosts a signal by 60db.

[/QUOTE]You could. No one does it because then you wouldn't be able to use it for line level signals.

The whole world settled on specific voltage references for line level, and that's what ADCs work with.[/QUOTE]

That makes sense. I didn't think about the need to go out of a DAW to line level machines during mixing and mastering. Still though, I suppose there could be a need for it on a consumer level. Makes me wonder if those USB mics that are out now have a preamp stage after the capsule or whether they convert at lower analog levels.

[karbomusic]

Originally Posted by buenavista

Changing the description would change the function because the function would be to convert an audio signal without the need for a preamp.

Where a mic signal starts and where it needs to get in order for you to hear in the real world is going to need to be amplified many, many times - doing that after the ADC doesn't fix anything.

[JSMastering]

Originally Posted by buenavista

I didn't think about the need to go out of a DAW to line level machines during mixing and mastering. Still though, I suppose there could be a need for it on a consumer level. Makes me wonder if those USB mics that are out now have a preamp stage after the capsule or whether they convert at lower analog levels.

So, you don't listen to music while you're making it?