Who mixes using their audio stems?
- Thread starter ALittleNightMusic
- Start date
quickbrownf0x
I had fun once. It was awful.
BOTH. But, I prefer to mix with stems in a new session. I like it, because it puts you in a different mode and it feels kind of nice to say 'okay, cool - now that composing's done, moving on!'. I give myself a cookie everytime that happens. 
Plus it saves a ton of resources and you only have to look at just a few (sort of) stems, instead of (in my case) a gazillion tracks. Keeps you focused.
Mind you, when I print my final stems, I think about 95% of the mix is already there, courtesy of my template and a whole lot of tweaking/cursing/crying/punching through a wall beforehand.
After that it's a bit of polish, tarting it up with way too much reverb and then exporting it to a
final mastering session. Final coat of polish there. Export/downmix to stereo (if it's not already). Upload/throw it over the fence. Bob. Uncle.
Here's an example btw, if you want to watch me twiddle my thumbs;
Plus it saves a ton of resources and you only have to look at just a few (sort of) stems, instead of (in my case) a gazillion tracks. Keeps you focused.Mind you, when I print my final stems, I think about 95% of the mix is already there, courtesy of my template and a whole lot of tweaking/cursing/crying/punching through a wall beforehand.
After that it's a bit of polish, tarting it up with way too much reverb and then exporting it to a
final mastering session. Final coat of polish there. Export/downmix to stereo (if it's not already). Upload/throw it over the fence. Bob. Uncle.
Here's an example btw, if you want to watch me twiddle my thumbs;
Last edited:
jbuhler
Senior Member
I feel if you haven’t printed to audio, you are at best delivering a rough mix. Of course a rough mix is perfectly adequate for many uses. And the vast majority of things I do for my own use never go beyond a rough mix.
Most of the rough mix is handled through orchestration, midi programming and automation, balancing and positioning instruments in the project, mic selection, reverb if needed, etc.
But I always feel it could use another layer of proper mixing using audio stems. And I’ve always opened a new audio only session in the few occasions I take it to a final mix.
Most of the rough mix is handled through orchestration, midi programming and automation, balancing and positioning instruments in the project, mic selection, reverb if needed, etc.
But I always feel it could use another layer of proper mixing using audio stems. And I’ve always opened a new audio only session in the few occasions I take it to a final mix.
Soundbed
Music for TV
Depends how many tracks (channels).
My larger hybrid orchestral stuff needs a new session.
Smaller (20 tracks or less?) is usually pop/electronic or drone / minimal underscore and I can do it in the sand session.
You didn’t ask, but…
My process is geared toward giving my publishers alternate mixes aka alts.
Regardless of any other bussing, I always have four busses before the master.
Bass, drums, backing and melody.
Anything interesting that would interfere with dialog goes into melody and gets muted for a no mel alt.
Drums (almost always) include all non pitched effects and some pitched fx like risers that contribute to a rhythm bed.
Bass includes synth bass pulses and anything that can be included in a “drum and bass” alt.
Backing is everything else.
Each alt needs to stand on its own and carry the same emotion as the full mix, without (many) silences or gaps. They all need to be independent “pieces” almost like you’d imagine a game mix where elements get muted and unmuted as the action increases or decreases. This changes the way we write and arrange; nothing can simply “hide” in the mix if it doesn’t also work in one of the four sub mixes.
Usually violas are in backing, but if they play melody I need two sets of violas.
Usually cellos are in the bass buss unless they are featured and would interfere with dialog, considered melody. Etc.
The whole session and arrangement (template) is geared for mixing before I even write a note. (For arrangement I usual have pre-saved markers for intro, stingout, and so on. Most cues are 2:00-2:30 depending on style and tempo.)
My master buss is relatively minimal.
The first is a simple tool to measure the incoming mix and adjust the whole level if needed. Then a light multiband compressor which I never adjust. Then a gentle EQ, which I try not to use. Then a limiter. Then a visualizer/measurement tool for lufs, stereo image, etc. Any big issues need to be addressed in whatever is feeding the four busses (not on the 2 channel mix).
(Most of my stuff is “home mastered” by me. A couple of my publishers master for us, and they require different stems delivered for their mastering engineers.)
One of my goals is to be able to hear basically the same sound if you listen to my full mix or mute and un mute the drum and bass, dnb + backing, and add melody sub mixes (Alts) … when I started out, I relied on the 2 channel mix to “remedy” a lot of problems in the mix. Now, my sub mixes add together to sound basically like the full mix, within a small margin.
It’s all designed for video editors to use modularly.
Although, in practice, they rarely take the time to do much more than throw any of the mixes that doesn’t have a melody, turn it down, and throw the sting on at the end.
My production music friends joke that we write the melodies for the music supervisors, so they throw our cues in the editor’s bin. But the editors only use the no melody versions (because I mostly write for reality shows, where there’s tons of dialog and nearly wall to wall music).
I’d be interested in hearing from @jcrosby on this.
My larger hybrid orchestral stuff needs a new session.
Smaller (20 tracks or less?) is usually pop/electronic or drone / minimal underscore and I can do it in the sand session.
You didn’t ask, but…
My process is geared toward giving my publishers alternate mixes aka alts.
Regardless of any other bussing, I always have four busses before the master.
Bass, drums, backing and melody.
Anything interesting that would interfere with dialog goes into melody and gets muted for a no mel alt.
Drums (almost always) include all non pitched effects and some pitched fx like risers that contribute to a rhythm bed.
Bass includes synth bass pulses and anything that can be included in a “drum and bass” alt.
Backing is everything else.
Each alt needs to stand on its own and carry the same emotion as the full mix, without (many) silences or gaps. They all need to be independent “pieces” almost like you’d imagine a game mix where elements get muted and unmuted as the action increases or decreases. This changes the way we write and arrange; nothing can simply “hide” in the mix if it doesn’t also work in one of the four sub mixes.
Usually violas are in backing, but if they play melody I need two sets of violas.
Usually cellos are in the bass buss unless they are featured and would interfere with dialog, considered melody. Etc.
The whole session and arrangement (template) is geared for mixing before I even write a note. (For arrangement I usual have pre-saved markers for intro, stingout, and so on. Most cues are 2:00-2:30 depending on style and tempo.)
My master buss is relatively minimal.
The first is a simple tool to measure the incoming mix and adjust the whole level if needed. Then a light multiband compressor which I never adjust. Then a gentle EQ, which I try not to use. Then a limiter. Then a visualizer/measurement tool for lufs, stereo image, etc. Any big issues need to be addressed in whatever is feeding the four busses (not on the 2 channel mix).
(Most of my stuff is “home mastered” by me. A couple of my publishers master for us, and they require different stems delivered for their mastering engineers.)
One of my goals is to be able to hear basically the same sound if you listen to my full mix or mute and un mute the drum and bass, dnb + backing, and add melody sub mixes (Alts) … when I started out, I relied on the 2 channel mix to “remedy” a lot of problems in the mix. Now, my sub mixes add together to sound basically like the full mix, within a small margin.
It’s all designed for video editors to use modularly.
Although, in practice, they rarely take the time to do much more than throw any of the mixes that doesn’t have a melody, turn it down, and throw the sting on at the end.
My production music friends joke that we write the melodies for the music supervisors, so they throw our cues in the editor’s bin. But the editors only use the no melody versions (because I mostly write for reality shows, where there’s tons of dialog and nearly wall to wall music).
I’d be interested in hearing from @jcrosby on this.
ALittleNightMusic
Professional Amateur
BOTH. But, I prefer to mix with stems in a new session. I like it, because it puts you in a different mode and it feels kind of nice to say 'okay, cool - now that composing's done, moving on!'. I give myself a cookie everytime that happens.
Mind you, when I print my final stems, I think about 95% of the mix is already there, courtesy of my template and a whole lot of tweaking/cursing/crying/punching through a wall beforehand.
After that it's a bit of polish, tarting it up with way too much reverb and then exporting it to a
final mastering session. Final coat of polish there. Export/downmix to stereo (if it's not already). Upload/throw it over the fence. Bob. Uncle.
Here's an example btw, if you want to watch me twiddle my thumbs;
Great mockup! And I agree that mixing is a different mode and you start to focus on different things than when you are composing / producing.
Bara Matahari Pagi
he/him
I usually mix on the composing project, because I'm such a perfectionist and often change some (composing-related) stuff even when mixing haha.
I also love the flexibility of being able to mix every single instrument individually, rather than having to mix stems.
Another plus is it doesn't take up too much storage space lol!
This is of course if the computer can handle it, but usually my PC works just fine
I also love the flexibility of being able to mix every single instrument individually, rather than having to mix stems.
Another plus is it doesn't take up too much storage space lol!
This is of course if the computer can handle it, but usually my PC works just fine
NekujaK
Atomic batteries to power! Turbines to speed!
Often bounce MIDI tracks to audio for mixing, mostly to optimize performance in large projects. But I still mix within the source project, simply because I can and it's convenient.
The flexibity to freely move back and forth between writing, recording, arranging, mixing, and even mastering, all within a single project is one of the advantages a DAW presents over the limitations of traditional studio workflows. However with all that power, comes responsibility. It really helps to be grounded in the fundamentals of each of these disciplines/functions to understand where the boundaries are, and how to approach them effectively.
The flexibity to freely move back and forth between writing, recording, arranging, mixing, and even mastering, all within a single project is one of the advantages a DAW presents over the limitations of traditional studio workflows. However with all that power, comes responsibility. It really helps to be grounded in the fundamentals of each of these disciplines/functions to understand where the boundaries are, and how to approach them effectively.
jcrosby
Senior Member
That's actually pretty similar to how I would bus things when I was doing only TV production music. I'd use a few more groups, but the same idea, and pretty simialr... Any percussion beds would also include pulses, (even braams or bass synths like 808s, etc), and SFX would also be part of the percussion bed. In addition to the main version I'd send a 'harmonic' bed, percussion bed, a version with no lead/melody, (plus cut downs, stingers, etc).
I've been focused only on trailers the last 3-4 years (mainly because it's a completely different mindset, also customization's always a possibility)... I plan on doing more general production music again when the imposter syndrome becomes more manageable
[True story!]). Now I typically have to send stems for everything, (minimum 12, maximum of 24). The stems also have to be an identical sum of the mix, meaning no bus processing, (although I cheat a little and use a little EQ and some Inflator), so I have discrete busses for everything, and all processing happens on the busses...There are my typical stem busses:
Hits & sub booms (May be bounced together or separate, depends on the cue)...
Drums/large percussion
High percussion (field recording percussion, 'tick tocks', sticks, etc).
Short strings
Long strings
Brass
Choir/vox
Top line/melody (if any... Trailers these days tend not to feature one)
Pads/atmospheres
'Signature' synths (if used)
Any other 'signature'/distinct sounds that an editor might want to feature, or mute.
Bass pulses
Mid/high pulses
Braams (or similar, i.e. 'reeses', distorted 808s, etc)
Lead synth (if used)
Short plucky synths (i.e. 'pings', 'sirens/alarms')
SFX
It depends on the project in terms of how many stems I send. If something is primarily orchestral or organic I may send 12 to 15 stems, if something's hybrid, and has a lot of sound design I may send between 15 and 20. (I believe the most I've sent for a single cue is 21).
As far as mixing and writing in the same session, that's a luxury I wish I had! The mix has to happen as part of the production for a few reasons...
A). Revisions are inevitable. Most cues have 3 or more rounds of revisions before being accepted, (3's the minimum I've received, and it's rare. More like 4 or 5).
When I get to those last one or two notes I'm typically fighting a lot of latency, (thankfully Logic has low latency mode).
B). 'Production value' is (more or less) expected to be part of your v1. Basically, if you wait to mix everything at the end you'll just get notes about the mix as part of every revision, so it's unavoidable. Fidelity can't be an afterthought with libraries who target trailers... And the fidelity really does have to be the best that you know how to deliver.
I do export stems, then bounce final mixes from there, but that's because cut downs are easier with stems, and more importantly, you never know what they'll come back with in terms of feedback until they sign off on it... So exporting stems before they say you're done would be redundant. The last note may say 'do these 2 or 3 things and we're all set', but I don't export stems until they say it's time...
Alt mixes are optional if the cue has no voice. Alts are required for cues with choir, or any other vocal. So a typical orchestral or hybrid cue with choir has two mixes as part of the deliverables, the full mix and a 'no choir' alt mix (as well as cut downs). If a cue does wind up having a distinct top line or melody I'll also bounce out an alt version with it muted and let the publisher decide if they want to include it as an alt mix, (which has worked in my favor a few times... An Aquaman trailer released a couple of weeks ago where they muted a brass/string top line I originally had in the cue, so it's not just reality TV editors that don't like melodies
)And although it may seem redundant having alt mixes when you include all of the stems, there is some degree of logic to that... As you said, it's all about making life as easy as possible for the editor. Alt mixes can be useful for targeting trailer houses that may be cutting a bunch of TV commercials for a specific campaign. Sometimes they'll use a cut down as is, or chop up an alt mix to fit a shorter cut...
I'll probably continue to work this way regardless of whenever I get back to writing some more general production music... I'd adapt and solo what I want as individual stems if needed, with trailers though, you have to be literal in how you group your instrument stems... Basically, my busses reflect everything I need to deliver as a stem because doing it any other way is problematic.
Although currently I have to deliver parts with no bus processing, for libraries where stems aren't required, and want me to master things myself then I'd do some moderate processing on the mix buss. A little saturation or tape, some mix bus compression or multiband compression, a little EQ, and some limiting. (Good old Inflator never hurts either).
And absolutely, metering is essential. With the publishers I'm currently working with everything's sent off for mastering, so meters really are critical.
Last edited:
As a hobbyist my goal with mixing is to get to sounds that make the process of composing as enjoyable as possible to me, which means sounding as good as it can - ideally without adding too much latency. So I mix directly in my midi/recording template but I might compromise on the master fx chain and disable that while composing to minimize latency.
vienthousiast
Active Member
I like everything clean, so I print the stems, mix a bit, re-print and once I go for a full mixing+mastering session (I always do them together), I do everything in a new session with the final stems. Nothing fancy but works for me since 10+ years now.
Soundbed
Music for TV
Thanks so much and huge congrats on the Aquaman trailer, that’s big time!That's actually pretty similar to how I would bus things when I was doing only TV production music. I'd use a few more groups, but the same idea, and pretty simialr... Any percussion beds would also include pulses, (even braams or bass synths like 808s, etc), and SFX would also be part of the percussion bed. In addition to the main version I'd send a 'harmonic' bed, percussion bed, a version with no lead/melody, (plus cut downs, stingers, etc).
I've been focused only on trailers the last 3-4 years (mainly because it's a completely different mindset, also customization's always a possibility)... I plan on doing more general production music again when the imposter syndrome becomes more manageable
There are my typical stem busses:
Hits & sub booms (May be bounced together or separate, depends on the cue)...
Drums/large percussion
High percussion (field recording percussion, 'tick tocks', sticks, etc).
Short strings
Long strings
Brass
Choir/vox
Top line/melody (if any... Trailers these days tend not to feature one)
Pads/atmospheres
'Signature' synths (if used)
Any other 'signature'/distinct sounds that an editor might want to feature, or mute.
Bass pulses
Mid/high pulses
Braams (or similar, i.e. 'reeses', distorted 808s, etc)
Lead synth (if used)
Short plucky synths (i.e. 'pings', 'sirens/alarms')
SFX
It depends on the project in terms of how many stems I send. If something is primarily orchestral or organic I may send 12 to 15 stems, if something's hybrid, and has a lot of sound design I may send between 15 and 20. (I believe the most I've sent for a single cue is 21).
As far as mixing and writing in the same session, that's a luxury I wish I had! The mix has to happen as part of the production for a few reasons...
A). Revisions are inevitable. Most cues have 3 or more rounds of revisions before being accepted, (3's the minimum I've received, and it's rare. More like 4 or 5).
When I get to those last one or two notes I'm typically fighting a lot of latency, (thankfully Logic has low latency mode).
B). 'Production value' is (more or less) expected to be part of your v1. Basically, if you wait to mix everything at the end you'll just get notes about the mix as part of every revision, so it's unavoidable. Fidelity can't be an afterthought with libraries who target trailers... And the fidelity really does have to be the best that you know how to deliver.
I do export stems, then bounce final mixes from there, but that's because cut downs are easier with stems, and more importantly, you never know what they'll come back with in terms of feedback until they sign off on it... So exporting stems before they say you're done would be redundant. The last note may say 'do these 2 or 3 things and we're all set', but I don't export stems until they say it's time...
Alt mixes are optional if the cue has no voice. Alts are required for cues with choir, or any other vocal. So a typical orchestral or hybrid cue with choir has two mixes as part of the deliverables, the full mix and a 'no choir' alt mix (as well as cut downs). If a cue does wind up having a distinct top line or melody I'll also bounce out an alt version with it muted and let the publisher decide if they want to include it as an alt mix, (which has worked in my favor a few times... An Aquaman trailer released a couple of weeks ago where they muted a brass/string top line I originally had in the cue, so it's not just reality TV editors that don't like melodies
And although it may seem redundant having alt mixes when you include all of the stems, there is some degree of logic to that... As you said, it's all about making life as easy as possible for the editor. Alt mixes can be useful for targeting trailer houses that may be cutting a bunch of TV commercials for a specific campaign. Sometimes they'll use a cut down as is, or chop up an alt mix to fit a shorter cut...
I'll probably continue to work this way regardless of whenever I get back to writing some more general production music... I'd adapt and solo what I want as individual stems if needed, with trailers though, you have to be literal in how you group your instrument stems... Basically, my busses reflect everything I need to deliver as a stem because doing it any other way is problematic.
Although currently I have to deliver parts with no bus processing, for libraries where stems aren't required, and want me to master things myself then I'd do some moderate processing on the mix buss. A little saturation or tape, some mix bus compression or multiband compression, a little EQ, and some limiting. (Good old Inflator never hurts either).
And absolutely, metering is essential. With the publishers I'm currently working with everything's sent off for mastering, so meters really are critical.
proggermusic
Keyboards and Woodwinds, Jazz-Funk-Ambient?
I hate to put on my "Adam Ruins Everything" hat, but I figure this is probably a good place to point out the differences between "stems" and individual audio tracks... they're not quite the same thing, although they're related, and the distinction might end up being very important to know for composers who are starting to work more with dedicated mixing engineers. I have no affiliation with or loyalty to Izotope, but they wrote a pretty good blog post about this topic:
Stems and Multitracks: What's the Difference?
Stems and Multitracks: What's the Difference?
ALittleNightMusic
Professional Amateur
I could've framed my question more ambiguously instead of specifying stems. In regards to that Izotope article, I just went through the Thinkspace Orchestral Mixing videos and Jake Jackson mixed a sample orchestra piece using a mixture of stems and individual multi-tracks. End result sounded pretty good! But I'm sure he would always want all multi-tracks if he had the choice.
Also, he had a very light touch compared to some other mixing videos I've seen for orchestral music. It's curious how little he did vs. how drastic some folks are (with like EQ cuts). These libraries are extremely well recorded and great sounding, so it has always been a question on my mind on why somebody is carving things up so aggressively (and whether the real issue lies in the orchestration and production).
jcrosby
Senior Member
You bet. And thanks, it's very much appreciated... BTW, I notice you come from an EDM background. Although originally a guitarist, I spent many years making EDM before doing production music, (mostly drum and bass), and did quite a lot of EDM & hip hop production cues prior to trying my hand at trailers... Having good EDM/electronic chops has proven to be really useful.Thanks so much and huge congrats on the Aquaman trailer, that’s big time!
Soundbed
Music for TV
Yeah for sure! I co-owned a micro label for deep house (primarily) with about 40-50 artists and we sent our mixes to a mastering engineer the LA area who taught me a bit both about my stuff and how he handled different artist mixes.
Some artists really expected their kick drum to sound noticeably more club ready after mastering, and were disappointed in the mastering which was amusing.
(Apologies for getting off topic.)
I have imposter syndrome for trailers, lol. I’ve got a couple trailer esque things that made tv ads but not a true trailer trailer haha. Mostly trailer inspired or trailer junior.
Beat Kaufmann
Senior Member
I've only been mixing audio files for years.
About 20 years ago, as a WIN user, I had to switch from Logic to Cubase. Today, all Logic projects are worthless, because back then I mixed everything from the project. Since then I play all instruments as audio files (without effects) and save my projects that way. Even if all the licenses and software were gone in 10 years, I could still remix my projects with the skills and tools of the future.
But there are other advantages too.
A) Already mentioned: ideal starting point for later mixes
B) If you mix out of the project, I've noticed that you quickly correct something on the midi side, on the audio side, then you master a bit again. When I separate composing/producing and mixing, everything is more efficient. I like to separate the individual work steps.
C) Mixing with audio files is new work. The project loads quickly. You can refocus on the individual instruments and give the individual voices a whole new level of support.
D) Before I start mixing the audio files, I normalize them all to the same average RMS level. This is a fantastic start for a mix with lots of instruments. The mix actually starts out very balanced this way.
E) I love "slim" projects. In the past, I often experienced that in large projects (samples + audio mix), for example, some effect plug-in was no longer accepted in a later DAW version or was even treated as corrupt. Cubase then often crashed, meaning that the project could no longer be loaded so that the unwanted part could have been deleted. Sometimes changed sample locations or missing licenses were also the reason for such a crash.
That's why today I like to open a new project for a next step, even though Cubase runs pretty stable today.
So much for how I work.
Beat
About 20 years ago, as a WIN user, I had to switch from Logic to Cubase. Today, all Logic projects are worthless, because back then I mixed everything from the project. Since then I play all instruments as audio files (without effects) and save my projects that way. Even if all the licenses and software were gone in 10 years, I could still remix my projects with the skills and tools of the future.
But there are other advantages too.
A) Already mentioned: ideal starting point for later mixes
B) If you mix out of the project, I've noticed that you quickly correct something on the midi side, on the audio side, then you master a bit again. When I separate composing/producing and mixing, everything is more efficient. I like to separate the individual work steps.
C) Mixing with audio files is new work. The project loads quickly. You can refocus on the individual instruments and give the individual voices a whole new level of support.
D) Before I start mixing the audio files, I normalize them all to the same average RMS level. This is a fantastic start for a mix with lots of instruments. The mix actually starts out very balanced this way.
E) I love "slim" projects. In the past, I often experienced that in large projects (samples + audio mix), for example, some effect plug-in was no longer accepted in a later DAW version or was even treated as corrupt. Cubase then often crashed, meaning that the project could no longer be loaded so that the unwanted part could have been deleted. Sometimes changed sample locations or missing licenses were also the reason for such a crash.
That's why today I like to open a new project for a next step, even though Cubase runs pretty stable today.
So much for how I work.
Beat
Jeremy Spencer
Senior Member
Not necessarily. You could just save them as MIDI and import them into Cubase. I’ve had to do this with old Sonar files.
jbuhler
Senior Member
This raises a different question, albeit off-topic: what do people find useful and convenient to archive for old projects? Do you bounce out audio for all tracks as well as your final version(s)? Do you export the midi? Or do you leave it all in the project, assuming that the old projects will remain accessible so long as you continue to update the software? Also do you disable all plugins before archiving to facilitate opening the project when plugins no longer load? What sort of notes to you leave to help reconstruct the project?
ALittleNightMusic
Professional Amateur
I think Beat’s point was that he didn’t bounce out stems and now doesn’t have the plugins / VIs originally used in the Logic projects.
