Gain staging. How? The basics

I have this massive orchestral project to gain stage/balance. By nature I am a musician not a tech - this comes hard to me.
I spent the last few days blowing my mind trying to understand the difference between dB and Loudness, trying to understand how to use the Cbase meters and trying understand how to understand how to preserve headroom. My mind is like Spaghetti all the threads of understanding lead nowhere. What I thought I knew, I no longer understand.

Here is my most basic question. I want to go through my project comparing flute to flute, trumpet to trumpet, and getting instruments from different manufacturers to be “appropriate” with each other. This means all the flutes to be broadly the same, but of lower volume than the trumpets. I also want to make sure I don’t max out at the output, either when doing this, or when the orchestra plays. The issue is, I really have no idea what dB or LU to aim for in the control room meters.
Some engineers suggest 6dB headroom. How do you achieve headroom? What does this mean in terms of your meters? In a practical sense, in Cubase? I heard one guy say that you gain add a gain of 6dB to the stereo output, then compose as “normal”, then at the end take the 6dB off leaving that headroom for masterers. This seems to make sense to me, but I know my knowledge is lacking.

Can anyone help with this basic question?

Use the pre gain in the eq to properly gain stage. Goal is to be able to have all tracks push approximately the same signal strength so nothing gets lost in the mix. After each track is staged you’ll want to utilize the loudness meter. I personally keep the mixer at 0db, and rout to a master vca. I bring the staged tracks to around 23 lufs. Then I export to a multi track mixdown. At 23 lufs you’ll have plenty of headroom for the mix process.

2 Likes

Sounds to me like you’re asking about mixing, not tracking… so I’ll stick with the former on this post. Draghe’s post reads to me like tracking—though I may be wrong, of course.

Have you read anything from Bob Katz on Loudness Wars? K-System Metering? There are different configurations which depend on the intended use of the final product—is it for broadcast, film, YouTube? All these have different requirements.

I’m a total geek so I started here: An Integrated Approach to Metering, Monitoring, and Levelling Practices — but you’re welcome to do your own research.

I have used the K-System for a long while now, and it has worked great for me. I set my meters to K-14 and mix so the (intended) loudest part of the track/album/project hits the zero mark. This leaves 14 dB of “headroom” which my mastering engineer can use later as needed.

I mix by section—strings, then woodwinds, then brass, etc.—making sure each section sounds balanced within itself, then use Group/DCA meters to balance the sections to each other. Watch the meters so loudest equals zero on the K-14 meter.

That’s all in a very, very small nutshell… Hope this helps?

Disclaimer

I am a composer/conductor who was taught mixing and mastering by my best friend, a Tonmeister. We both worked in the film industry. There are peers on this forum who are certainly much more qualified to guide you than I am, so please do not take the above as a definitive answer.

1 Like

I am definitely talking of balancing a template. The template itself will not be used to compose, tracks will be drawn from it into other projects . It’s for the stage, so I don’t need and don’t want a compressed sound. I want a wide range of dynamics, from ppp to fff, if I can achieve this. I really do not like this “loud is best” approach.

“I mix by section—strings, then woodwinds, then brass, etc.—making sure each section sounds balanced within itself, then use Group/DCA meters to balance the sections to each other. Watch the meters so loudest equals zero on the K-14 meter.”
Yes but how…? I have, for example, fourteen flutes East West is Loud and brash, Orchestral Tool is almost inaudible.

Thank you for the valuable reply. A couple of questions. I think in ppp to fff not in dB or LU. I just have a smattering of what these mean.
If I am listening to say a flute sampled at pp, the ‘proper’ level of that sample depends on two things outside of Cubase. First the dB level baked into the original sample, second the level of my amplifier (currently set at -25dB). How do these two factors affect your end result? I think it might be possible to balance the sounds all wrong simply because you have your amp set wrong.

Z

Here are the things I have done already.

1] I have gone through my entire, very sizable library and added every usable instrument.
2] I have folderized and colorized every track according to the following hierarchy, each folder gets its own ‘subgroup’ which then feed into further subgroups and then Master Groups which feed to the Stereo Out.
3] So, for example, Violins are grouped/folderized into Solo Violins and Ensemble Violins. These are within a Master Folder called Strings. The individual instruments (all key switched with expression maps where called for) route into the subgroups, Solo Violins and Violin Ensembles. These route on to the Strings Master Group, which then routes to the Master Orchestral Group then Stereo out.
4] I have written four bar scale run within the range of each instrument, and set their mod wheel values for 100 and their velocity values for 64 (mf I hope) . I have used a key switch direction to select the legato patch or failing that a sustain patch. I have copied this part to every instrument in the same class. I have done this for comparison purposes.

5] I think I shall add a master VCA, as Draghe suggests.

The basic overall structure of this project is:

Strings
Brass
Woodwinds
Percussion
Harp
Pianos
Guitars
Synths
World

What next?

Now is time to utilizing the loudness meter. 23 lufs avg is the goal. To do so use the master vca I suggested, linked to the group track faders you created, and get a lufs level of 23 average. Export audio mixdown to 32 bit float multiple tracks. Choose create new project and save in project mixdown folder. Lable Mix version 1. It should export and open the new mixdown project. You’ll now be able to mix your instruments with the headroom you desire. You may choose to do so without creating group tracks, and then create group traps within your mix to leave more headroom for individual instruments. Note: This will work once your tracks are all rendered from VST instruments to audio if you have any in your project. Once it is in 32-bit float environment it’s pretty much unlimited headroom. Now is just a matter of getting your loudness radio-friendly. Which today means stream friendly. You want to bring it up to around 18lufs. If you’re doing your own mastering you can go right to 14 lufs. That’s what streaming sites generally use. You may want to master multiple versions at 14 16 18 etc. The db level does not matter at this point. In the digital world you’ll hear distortion and clipping. As well as the elimination of headroom. that is why keeping the db level at -14 to -6 is welcome to orchestral production. A violin will sound louder than a bass at same db. This is the concept of db vs loudness.

3 Likes

With respect. I am not looking to write for media or steaming, its for theatre. I am looking for a wide dynamic range not the compressed “loudness wars” sound. There are over 1600 tracks in this template, so rewriting it is not an option.
Honestly I only partly understand your answer.
The VCA fader thing. I have never used one. I will read up on it. I loaded one and found no fader in the fader - weird. I shall do my homework tomorrow.
The groups. This is the plan. Use the groups for futher mixing and levels in the compositional pieces. Have a blank template with only the groups in it. Audition sounds from the Master template then “Import into project” the required instruments.
This is (mostly) classical music, it requires a different approach to content for streaming, it suffers greatly when the dynamic range is decreased.

Thank you for your thoughtful response

Z

VCA and DCA are essentially the same thing, so don’t stress about the name… In Cubase, VCAs control multiple faders in a group of faders, such as in the structure you described. Your structure sounds good; and yes, you can add a master VCA as Draghe suggested, it won’t hurt unless you turn it down to infinity, haha!

For stage, you probably want K-20, same as film (i.e. in a movie theater). There is very large dynamic range in this (K-)system.

For a true, accurate picture of loudness, you’ll want to calibrate your speakers. Add a track with a pink noise generator at -20dB and run the noise through each speaker separately, one at a time. With a dB meter set to slow/C weighting at your listening position, change the amplification so the meter hits 85 dB for that speaker. Once both are calibrated as such, you shall be very close to the specs without too much fuss.*

Next, go through each sample of your instruments and change their individual gain (clip gain works great if they are audio files) so they have the same loudness, e.g. all flutes at ppp or fff will sound ppp or fff, respectively. I’d start with the fff to minimize clipping, etc. You’ll have to work instrument by instrument, especially if they are coming from separate libraries.

Then, you can mix the instruments in sections (e.g. all strings); balance the sections (e.g. strings balanced to woodwinds), and so on, via their DCA/Group faders. Get a rough mix going (large corrections), then finalize (small corrections) after you have the actual musical material in place.

Hope this helps!

*This is what the text atop the K-System picture represents.

Reading your inital post, I guess it would be good for you to take a day or so to learn something about decibel, loudness and how the human ear works. There are some nice videos on YouTube explaining different aspects.
Most important aspects maybe are:

  • ppp to fff (as anything else on a score sheet) are instructions to be interpreted by the artist. Decibel and LU (Loudness Units) are absolute values, not to be interpreted by an artist.

  • dB refers to the power (energy level) of sound (how violent are those air molecules moving) whereas loudness refers to perception of the ear/brain system

  • Inside a mixing console or inside Cubase, dB means the difference of a level (whatever level the dB is referring to such as dB(V), dB(U), dBFS). So zero dB = no change in level.

  • Headroom is the difference of your signal level and the maximum possible signal level.

If I understand you correctly your instruments from libraries contain e.g. 20 different trumpets each of them having a different volume.
I would create 20 tracks for the trumpets and a folder track to put them all inside. A folder track is only for display organization and does not influence the sound unless you use something like its mute button. Initially all volume faders should be at 0.
Then I would create a group track and route all trumpet channels to that group. From your description that is basically what you have done already.
Then I would start to have something played at the maximum volume from each of the trumpets (that volume comes from the VSTi you are using). The channel with the quietest trumpet should be your reference channel. Set it so that its peak display does not go beyond 0dB.
Now you can go through each of your others channels one by one and compare its displayed level with that of your reference trumpet. Lower the volume fader to match the reference volume or even better, do it by ear.
Once you have done the channel by channel settings, you can play back the entire trumpet set and do small corrections to the levels, again by ear.
Later if you want to use the trumpet ensemble in a song you can adjust the ensemble volume by the volume fader of the group channel.
In case you want to work with plugins on the individual track channels, I would use a VCA fader after levelling the entire trumpet section. The VCA fader would be linked to all indivual trumpet channels and can reduce their volume evenly to give the plugins some room to work with. Let us know if you need to work with insert plugins on individual instruments.

This example was for one set of instruments only and needs to be repeated for each set you use. Basically you level your orchestra. Something the conductor would do in reality.
In the end you would use the volume faders of your various group channels to select , how loud each set of instruments should be.

If you want to know to which level you need to mix your final output, just ask whoever you are working for what they require and set it up in Cubase.

Reading your initial post, I guess it would be good for you to take a day or so to learn something about decibel, loudness and how the human ear works. There are some nice videos on YouTube explaining different aspects.
Most important aspects maybe are:
Yes, done this. There are a few things that throw me, but I shall leave these issues to another post. I agree with your summary. I understand that both MIDI and dB are logarithmic scales, that LU and dB come in different flavors, that dB is a relative scale (and is often not cited as such) . That there are momentary or peak values, longer term values etc. Under the hood, dB in particular is fiendishly complicated - but let’s not go there. in this thread please.

  • Headroom is the difference of your signal level and the maximum possible signal level.*

What I think I have found out about headroom is this:

  • That it is not shown on Cubase faders.
  • That it matters less in 32 bit sound engines.
  • That engineers like at least 6 dB for mastering.
  • That it is important to preserve partials, in say cymbal hits -without it, nasty clipping occurs

On dB: One YouTube engineer suggest adding 6 dB to the Stereo out, then composing as normal, then before going to the mastering process, take it off again - viola, 6 dB of headroom. Logically, I can’t see anything wrong with this but I hesitant to employ the technique as I can’t expertly evaluate it.

If I understand you correctly your instruments from libraries contain e.g. 20 different trumpets each of them having a different volume.
Correct. They widely differ in volume even when playing the same articulation. Can be as much as 30 dB!

I would create 20 tracks for the trumpets and a folder track to put them all inside. A folder track is only for display organization and does not influence the sound unless you use something like its mute button. Initially all volume faders should be at 0.
Then I would create a group track and route all trumpet channels to that group. From your description that is basically what you have done already.
Correct.

Then I would start to have something played at the maximum volume from each of the trumpets (that volume comes from the VSTi you are using).

This is what I have done - which differs a little. I think this also works. Your opinion?

I have created a scalar run for each instrument within its range. Mod Wheel Initialized for 100, velocity at 64 steady and bland. I have left expression alone. Volume CC is . I am using the master key switch instruments for each instrument. I have written all the expression maps. In the Key Editor I have tried to select the same patch, or near as similar - legatos, sustains non vibrato. I have no plugins in use at the moment. I am working one instance of Kontakt/Halion/Play (henceforth termed ‘player’) per instrument.

Microphones

I note that different libraries have different microphone set ups. Cheap libraries have only one microphone set up, better libraries have up to seven -some of them trees. I feel on the horns of a dilemma with this, because where there are more microphones available, I will use them, but this is not comparing like with like. Of course this effects volume balance.

The channel with the quietest trumpet should be your reference channel. Set it so that its peak display does not go beyond 0dB.
Now you can go through each of your others channels one by one and compare its displayed level with that of your reference trumpet. Lower the volume fader to match the reference volume or even better, do it by ear.
Once you have done the channel by channel settings, you can play back the entire trumpet set and do small corrections to the levels, again by ear.
Later if you want to use the trumpet ensemble in a song you can adjust the ensemble volume by the volume fader of the group channel.
In case you want to work with plugins on the individual track channels, I would use a VCA fader after leveling the entire trumpet section. The VCA fader would be linked to all individual trumpet channels and can reduce their volume evenly to give the plugins some room to work with. Let us know if you need to work with insert plugins on individual instruments.

This seems like good advice. I am still digesting it. Today I plan to investigate VCA faders and try to understand their role. I hope you understand that I need to look at this deeply, as there are 1600 tracks in the Master Template. I don’t want to mess things up and have to unpick it all. I have already has to do this once - unpicking is messy. I need to be sure.

I have a few questions re this paragraph please:

How to adjust gain? My logic is like this. First the sample(s) have their own baked in dB, then (pre-fader) this can be adjusted by:

The volume in the instance of the instrument
The Master Volume of the player
Cubase’s pregain

Does it damage a signal to max out a digital volume - if there is no clipping?
My instinct is to use the volumes in the following order. Instance, Player master, Cubase pre-gain.

I think this makes no difference to headroom? Do these adjustments make any difference to headroom pre-fader? This question makes my head spin and stops me from proceeding.

This example was for one set of instruments only and needs to be repeated for each set you use. Basically you level your orchestra. Something the conductor would do in reality.
In the end you would use the volume faders of your various group channels to select , how loud each set of instruments should be.
I think my answer to my own question about this, would be that it would take months to try to balance each articulation, and anyway, when composing there is room for further adjustment.

Each instrument within a section, and each section in comparison to other sections have widely different volumes and timbres.

Of interest:
I did some close studies of two piccolo samples. One from Orchestral Tools and one from East West Quantum Symphonic Library. Out of the box, on a level playing field, the OT piccolo was only just audible (yes really!) whilst the East West Piccolo was robust and of a proper volume. I think the difference was approaching 30 dB.
I then adjusted the gain and used Cubase’s excellent new plugin Supervision, using both trhe loudness meter and the dB meter. I got them to read the same for both plugins. However, my ears told me a different story. They did NOT sound like they were of the same level. I think this was due to the OT piccolo sounding thin and shrill, whilst the EW piccolo had more body, was more fruity.

From this experiment I learnt that it is your ears, rather than even the most accurate LU or dB meter, that must be the final arbiter.

Do you agree? I think you do.

A few further observations on this.
Each instrument is a key switched instrument, capable of many articulations. Different articulations sound at different volumes. The classic example is the difference between pizzicato and arco - arco can be fff but pizzicato can never achieve this. Other instruments have their own idiosyncrasies in this respect

Thank you for your sound advice, I look forward to your reply.

thiagotiberio

Thiagotiberio: Thank you for that. Presently I have no idea what K20 is, but I shall find out. I will calibrate the speakers in my studio again. I will call in an engineer (probably) .

That is a great idea. When I was building my studio, even after months of research and consultation with friends who really know their stuff, I called in an acoustician. In half an hour, I learned quite a bit more!

I have got a lot further with these concepts now. I think I’m ready. Thank you to all in this thread, Especially to those that introduced the concept of VCA faders and Bob Katz K theory.

This is my plan, given the work I have done already to get the Master Template Working.

I am going to use Voxengo Span on my Stereo Out.

Here is how:
Most advice on the internet about gain staging is inappropriate to my needs. I do not need to compress my audio for maximum loudness, as is the case for anyone wishing to join the streaming “Loudness” wars. This is an orchestral template . For orchestration we need the maximum dynamic range/ headroom, not the maximum overall loudness. Ppp to FFF
The Bob Katz System defines this Classical range as K20. Voxengo Span can select this option for it’s metering.(bottom left) . K20 means metering to -20dBFS. Lots of headroom .

So, after thought I see two tasks. Gain Staging and Balancing. What’s the difference?

Gain staging here means setting two or more instruments to near identical dB levels. This is appropriate for a folder full identical instruments - say classical flutes, or some other specific instrument. So gain staging is the way to go here.

Secondly, all instruments within a section need to play at appropriate levels to each other. I do not mean here - the same level. A clarinet will be uniformly louder than a flute where both are playing mf as marked on a score. This is what I term “balancing”. Another term I am using to myself is “parity” . This is where one type of instrument is appropriately louder or softer than it’s companions.
One danger to avoid is mission creep. By this I mean it’s impossible to balance all instruments in one session. It’s all too easy to keep ramping up the volume, little by little, as one works through the folders. This is wear metering steps in. It gives an objective foothold in the real world.

So the first step is to gain stage identical instruments. Of course we do not want to use any faders. We save the faders for compositional purposes. There are three possibilities:

We can set the Voxego Span plugin on the master out, then compare the readings of two identical type instruments against each other - starting with the quietest instrument at full velocity. For this I loop four bars of a scale within the range of the instrument. Using a duplicate of this track for all piccolos etc.
There are three places, pre-fader, one can adjust the gain. Firstly in the instance of the instrument (I am working with one instance of the player e.g. Kontakt. per instrument). Secondly, in the master volume of the Player. and thirdly the pre-gain of Cubase. I have opted for the pre-gain in Cubase as this is easy to see in Cubase’s mixer.
So, in pairs proceed by comparing all identical instruments in a folder.
When this process is complete, make a note of the level shown consistently in Span, then go through again and make final adjustments by ear. Why? I have found that even when the machine exactly matches the dB of two samples, to the ear they may not sound the same. This is because some instruments (piccolos for example) may be shrill and thin, whilst other piccolos may be throaty and more robust whilst being perceived as of equal volume. The human ear must be the final arbiter.

Go through all the instrument folders using this process. Once you have finished all identical instrument from any manufacturer, must sound near equally loud.

Next comes the balancing - first instruments within a section, then section against section. Here one may use the VCA faders, the group channel faders, or one can additionally adjust the pre-gain. The latter is probably preferable, (Because it is least likely to be used when actually composing) but a little messy to implement, as this is the second visit to the pre-gain, though one could use linking. The end result should be the sectioned are balanced and the instruments within them. I am still thinking this through a bit.
Lastly, it is also necessary to use EQ to tail off unwanted lower frequencies. to prevent muddy mixes

What do you think of this master plan people and peoplettes and non binary in betweenies?

Z