Recently i've been getting into a workflow in which from a base idea I aim to consolidate groups of related audio tracks, namely "drum bus", "bass bus", "vocal bus", "synth bus", etc... And then get every exported .wav from those buses back together and process them in a final "mix bus".

Don't know if anyone here have worked that way or similar (maybe it's a standard), that can point out things that one should consider when taking that approach, also some pros or cons of this practice. Right now I think it's good to lessen the workload of the cpu, and on the other hand you use more of pc storage in a single project. Also do you reccomend to use a limiter on each bus?

 

You are basically describing stem mixing. It's about as close to a "standard" as you are going to find, but that's not to say everyone does it. Especially with faster and faster cpus we can get now. You can leave everything unbounced almost to the end of your productions if you want to.

It's a very good way of storing backup projects for some people, and some people prefer to bounce every track individually. But the result is that you will be able to get that mix/track back open in the future, because all the plugins and effects are bounced in those files. You are not depending on having the same DAW or plugins, so some people consider it "future-proofing".

You can use a limiter anywhere you want, and have it doing a lot of work limiting; or almost nothing at all, depending on how you set the threshold of the limiter. If you are really trying for "competitive loudness", that limiter is more commonly becoming a clipper loaded there instead.

 

I've done mixes for folks that sent me tracks like that but I don't do it with my own stuff. Nowadays you don't really need to worry about it CPU-wise and I like the freedom of being able to 'fix' anything immediately. My projects are half MIDI tracks triggering 16 sampler channels and some synths, and the other half are software samplers/synths. After I have the arrangement down, I record the MIDI stuff as audio and then render all the software instruments into tracks. Then I save a new version, the 'mixdown' copy, delete all the MIDI/software tracks and work with the audio direct for the final product.. I can always go back to the original and fuck with it any way I want later, but I have the mix session itself for whatever copy I put up on Soundcloud/Bandcamp/etc to recall on its own.

Plus, now I've got all my audio backed up from bar 1 as individual tracks, so worst case scenario, I can restart the mix from scratch in a new DAW or whatever.

 

I started mixing like that after trying a top down mixing approach that ultimately didn't work for me. It goes from individual tracks to group busses (like Vocals, Guitars, Bass, Drums, Keys, etc) and then into a mix bus. I don't export anything, it just routes like that. I personally don't like to lock my tracks, since I very often go back to some individual instruments and busses for minor adjustments. I usually only accept to work on stems fully mixed and locked in if I'm mastering, when mixing I like to keep it open.

 

The biggest con really is the inability to work on individual tracks (obviously). Track processing, editing, panning, parallel processing, automation, you need to have all of that figured out before exporting the buses. And if you have something that belongs on one bus interacting with something on another bus (like sidechain compression or some triggering thing) then that is going to also be an issue if you decide to re-ballance the buses on the mixbus session.

So depending on how important that is in your case it might be viable, but you need to be well aware of all that or else you'll find yourself re opening the original session over and over again.

Now, if CPU is a problem, then maybe consider freezing some tracks or buses instead of exporting them and working on a different session. Many DAWs will also offer ways of organizing your tracks so the project doesn't look too cluttered visually, so you can pretend the individual tracks are not there until and if you need them.

 

It could be the case that the "stem mixing" was the more intuitive way to do things back in the days when the cpus couldn't handle entire projects, and now that they can, the "all at once" way is getting more and more common. I was working that way too, I used to have enough cpu for what I needed but I'm getting into more complex schemes (well also my m1 mac is getting old), like processing the low and high end of the bass guitar in diferent tracks and even getting the high end in two tracks so I can pan them each one in different sides, that's a lot more workload for the cpu. Maybe my problem lies in project organization like mk says and it could solved just bouncing and freezing tracks, things that I don't use to do, and yeah I was fearing that the project could get clustered, but maybe it would be the best to give that a shot. Thanks a lot for the tips

 

Top-down mixing is fine if you like it, but working with printed group stems isn't optimal for that approach. You are still diving back into the individual channels for any little adjustment or thing you have to fix. Personally, I would rather print each channel separately before mixing and bouncing them as stems of each bus. That way, they are there if I need them. Also, since I use Logic, there is a good timesaver "Bounce and Replace All Tracks". It reminds you to make a backup copy during that process.

Even if i do bounce everything as group stems, I take the stems and drag them all up to the top of the project, and then I mute and hide everything else. Disable every plugin on all the other channels. You will not even notice them there, in terms of cpu use. In Logic, you can select all your channels, hit X to load the mixer, and then disable the effects on every channel in that insert slot position. So you are clicking 1 button instead of 1 x your entire channel count.

Sometimes duplicating parts is the right way to go, but if you are just doing it to frequency split; you should try Blue Cat's MB-7 mixer too. It's a multi-band mixer where you can load in effects plugins to the individual bands, like Patchwork basically. There are others that do the same basic things.

 

For me, commitment. Working fast is important not only because 'life is short' or because 'we all have lives beyond work', but also and perhaps more importantly because many of your first and immediate ideas, decisions and actions are the best you'll have. Don't get me wrong, but I hear the sound differently after it gets bounced. Also, you can always change and revert the change on that reverb tail automation over and over again, deciding which connections are more relevant — is it the theme of the lyrics, which at that exact moment touch the concept of lingering, or is it the vocal performance, which lacks sustain at that point, or is it the interplay with the cymbals and so on and so forth... and does that really matter more than the other dozens of arrangement decisions you have on your to-do list? Maybe you should even toss your previous programmed tracks after bouncing to find real commitment. Live performances can't be recalled note for note, mod wheel bezier curve point for point, automation envelope on every knob etc. Bouncing restores the healthy sharp line that separates the realm of ideas from the realm of concrete action, a line that has been blurred by the ability to recall anything (mostly MIDI).

 

I did this only once in 2003 because I was bringing my 750MHz PIII PC to it's knees, so I rendered the busses to WAV stems, loaded those stems into a new project, and added recorded a few more tracks like backing aux percussion, etc. and very minimal leveling. Then, applied some light limiting on those stem files, bounced it down, loaded it into WAVELAB and did the DIY mastering on it.

Bottom line for me was that I got an interesting result, not bad or better, but a different 'rendered color. But had to do it because of CPU load, and probably HDD throughput. Back then you'd have to commit and destructively edit (more often, individual tracks) which was a really helpful lesson.

 

If you take the computer part out of your example this is the traditional way of console mixing mate. Grouping related tracks on a console's busses that may share identical "treatment" as a whole, regardless of the individual insert or send effects that you have already established. With the exception that the individual tracks are still accessible.
Going to the computer part now, exporting those groups into stems and arriving to something like -let's say- 6 or 8 final stereo tracks, is how lots of studios and engineers actually master nowadays, as it gives many more "surgical" options to the master engineer, than a stereo mix file.
But for mixing, i would still want the freedom of fine tuning individual elements until the very last moment. Exporting those stems is after all a destructive process and going back to the previous state of the mix means re-doing the whole thing again, so that is not very practical. Keeping everything up and ready to be tweaked at any moment is the best to my experience, no matter how many channel groups are involved in the mix. Everything is volatile and under constant consideration during a mixdown. Even tracks you already "finished" soundwise early in the mix process, may be altered after a radical decision that may affect anything from the arrangement/orchestration all the way to individual fx processing etc. And these decisions are possible at any moment.
Mind you, i am always referring to a professional environment, where the artist and the engineer are not the same person. And musicians are one of the most insecure "species" even when working with someone they completely trust. It's not the engineer, it's the artistic approach that -theoretically- implies any mix can always be improved hehe. Of course if you follow this to the letter you 'll end up mixing the same song for years haha. Just remember art is always unfinished and you 'll sleep better at night. The most important thing is to be able to move forward with songs,mixes etc. Enriching your artistic output and moving on to the next thing. Also remember "unfinished" does not by any means imply sloppy...
Personally, it took me years to differentiate/distance myself from my artistic persona when mixing my own tracks. I had to wear the technician's/craftsman hat but still maintain my artistic creativeness, something that i struggled with for a long time in the beginning, many times ending up never being satisfied with the outcome. For some peeps it just never happens, that's why they prefer others to mix their tracks for them. I do consider this a "healthy" approach, as i do admire artists with a strong self awareness, something which i find a rare trait.
Cheers
PS: When a mix won't "budge", just leave it for the next day, and approach it with a fresh ear. Ear and brain fatigue happens even within the best possible audio environment. So taking breaks can be proven just what you needed to finish off a mix and move on.

 

Last edited: Dec 29, 2024

Sine: You don't need a limiter on every bus. It may be useful on a drum or distorted guitar bus, but can also affect transients. I only put a limiter on the master, mainly to lock in my max volume limit. Before that, I do diligent level mixing per track, including automations for transitions.

Cosine: I think you have the right idea regarding bussing out track types. I would also recommend committing to audio on an individual track level when you feel confident in the production. It takes more storage, but ensures you lock in all that precision work.

Tangent: I work in Ableton, so I have the lucky convenience of keeping VSTs as midi there, then loading the project in Bitwig for stemming (because I don't have to make duplicate tracks). Having both midi and audio sessions is ideal.

 

For me it's a 3 stage process, where I create buses in the middle of the mixing process:

1) Make a MIDI mix with all the MIDI instruments and rendering to WAVs along the way when I think the tracks sound alright.
2) Turning off MIDI tracks and mixing WAVs into a pre-final mix, creating buses along the way, rendering buses/stems when I think it sounds alright.
3) Turning off WAV tracks and mixing buses/stems into a final mix.

This leaves you with all the options open, an easy way to change whatever you want later on, and you get a WAV backup of everything.

 

In Logic I combine all of these elements.

I create Sub Buses for sounds I think should live together. EG Drum / Guitars / Vocals /Percussion... I try to leave each bus as untouched as posible. Maybe add limiter or compression and touch of EQ. I can always switch these later if they're minimal. I try to leave the bussed-groups at 0DB and then mix each of the channels within the bus as if it was going to master. The latter happens before the former. Finally, I can then send the sub-group bus to FX buses. EG Reverb.

The main advantages I have found from this method
1 - Easy to make changes and A + B different ideas.
2 - Fine control of each channel.
3 - Easier to listen and mix sub-sections. EG Drums, or when mixing drums and bass guitar for example.
4 - Simplifies FX options. EG sending all rhythm guitars, or drums to same reverb bus.
5 - Simplifies use of outboard gear.
6 - See 2, can always add FX or mix at a fine grain level

In summary, grouping the channels into sub-busses enables a point of interception which enables some useful tricks and opportinuties for experimenting. At it's simplest it makes it easier to isolate groups of tracks. Sure, you can mute or solo the relevant tracks, but if I have 50+ tracks it's annoying to go thru individually to just hear the drums or guitars or vocals. However, there are many really useful cases for FX on buses and FX buses.

 

Since I play everything live on each individual track, guitars, keys, drums, vocals, etc, there's no need for freezing tracks. I do make busses for background vocals, drums and percussion, and a separate cymbal buss which I'll then meld/create a buss with the drum/percussion buss. I generally glue each buss with a compressor (individual tracks already have compression as I record with either a Warm WA76 or Audioscape Opto lightly mostly for flavor) and perhaps their own reverb.

 

Very valuable insights on mixing you have shared here, I think that with those tips I can fine tune my workflow, thanks a lot everyone

There's just one more thing that I would like to address on this topic. Some of you said that when you bounce a track some kind of colouration occurs, I felt that too when exporting mixes. That colouration depends on the hardware you own, the interface, right? It would be the case that the more tracks you bounce in a project the more of that colouration you get in your mix? Of course that colouration could be a good or bad thing, but it's important to make sound-wise decisions

 

Nope. It is the summing algorithm of your daw along with what plugins you have chosen to put in those tracks you exported.

Errr nope. It always depends on the genre of course. If you wanna mimic early Skrillex shit then perhaps brickwalling everything might get you there hehe. Personally i find this kinda terrible from a funny perspective. But i'm too much of a purist when it comes to processing individual channels. My (oldschool) motto is simple, if the track you played or recorded doesn't meet your expectations then you must redo it and not try to fix it in the mix.
I try to use as less compression as possible. If i can get a proper balance just by automating volumes, i will happily do so. Using comps to achieve a certain sound though, like sidechaining for example, is different and it belongs to the fx processing domain rather than individual "surgical" dynamic processing.
I'm also not a fan of any plug or even top notch dynamics outboard inserted on my mains buss while mixing. It just takes away the clear picture of a mix and replaces it with a pseudo, semi mastered or leveled stereo output that doesn't represent the actual true state of the mix. But this is my own approach, it doesn't imply everyone else's is wrong. It's what works for me actually. If you can get a great sounding mix with a whole chain of plugs in your mixbuss from the beginning, then it 's fine by me, since there are truly no rights and wrongs, only different "approaches".
After all we all want the same for our mixes, to sound as good as possible.
I try to create sounds as close to perfect as i imagine them to be from the early creation stage. If i am recording a band, i will spend a significant amount of hours in finding the correct miking for amps, drums etc, just like i was taught at an early age. Getting the best possible raw sounds and audio signals before you start mixing, no matter the genre, be it a great sound from a synth or incoming audio from a mic etc., is to me the most fundamental part of practically realizing/putting together a song.
All the points i referred to, make a reasonable entry as to why albums like Sade's Diamond Life are to this day considered, mix reference albums. Recorded in late '83 at London's Power Plant, you can clearly hear the band's dynamics, sultry atmospherics and soulful sounds in the best possible way. And no matter if you like the style, any critical mind would find the details and nuances in this album exquisitely portrayed. Why ? No compression heheh. Just a tight ass band with some imaginative soundscapes.
An opposite example should be Metallica's Death Magnetic, which no matter the Billboard #1 and Rick Rubin's signature, sounds as flat as a perfect square wave. Even the 2015 re-master is still bad lol. I actually heard some of the album's songs live attending a Metallica concert and they sounded far better than the album does.
Cheers

 

Last edited: Dec 29, 2024

Absolutely, without a doubt and without exception!

 

Why do we export to wavs & then arrange them again in our daws in this day & age! I only use the vocals as a wav file!I know it saves on the cpu etc.. but it seems too much work twice.. & if the synths/bass/ keys are low on the cpu, you can do it in one take not re do it again with wavs. Sounds like the old days when we used Cool edit pro or acid pro just to cut & paste wave files..

 

Last edited: Jan 1, 2025

A couple overlooked functions in Logic are great for this kind of thing. An example is using region clip gain and also the gain utility plugin, so that every fader can be reset to unity using option+click a channel fader with all tracks selected. I see people doing this more often in Ableton. I think they may not realize that with all tracks selected, any value that you manually input into the Region Inspector on one track will apply to all regions on the selected tracks. Every region can have it's own settings on all individual region parameters.

Another one that most people do not use because they don't know what it actually does is Normalize Region Gain. They think it is regular old normalization.

These are great for gain staging purposes, and especially at the start of mixing someone else's Logic project. The Summing Track Stacks are very useful if you want to do top-down mixing. You set the sum to your target for that group, and then go into the individual channels and basically work backwards while maintaining the same peak level for the whole stack. It all enables you to easily return any fader you want back to unity. If you use Split At Playhead, any section of audio can become its own Region in 1 key command; which lets new properties to be set for that individual region you just chopped.

 

Last edited: Dec 29, 2024

I don't like freezing or exporting busses because they play a very crucial role on my template. Especially when you got tons of parallel processing happening from the insert channels straight to the 2buss. I don't get any trouble with CPU those times, but im always freezing VSTi/MIDI instruments to audio, and sometimes some mixing channels and processing (like reverbs, automation) to get more control.

I think it like a tape machine. On the analog days people would die to have the amount of tracks that we can have in the box today with this ease. You can end up have a huge amount of tracks in a way you couldn't have on the tape. Quite honestly, i feel blessed