I just finished a mix for a client and was preparing for mastering, when it occurred to me that I get a lot of questions about how to make proper stems. I thought it would make a good topic, so here we go!
Here is the typical list of questions regarding making proper stems for mastering:
Should I use the Buss FX or not?Which stems should I make? How many is too many?Should the stems have reverb and delay or should those be separate stems?What sample rate and bit depth should I use to make the stems?Do I need to dither the stems?Should I use stereo or mono tracks?Do I bounce from zero at session start? Leave space for reverb/delay tails at the end?How do you preserve the side-chaining?How do you deal with stems from parallel busses?
The first question you need to ask yourself is:
I. Are these stems for Mastering or for Performance?
The answer to that question will inform how you proceed.
If the stems are for mastering, then your goal is to break a complex mix down into composite parts, that when combined at unity, comes extremely close to representing your mix.
If the stems are for performance, then your goal is to create a "mastered" sound for all of the stems, because they will not be summed through a master buss with FX.
In other words, the performance stems should include the master buss chain, so that each stem has power in the live setting.
For the discussion today, let's focus on preparing stems for mastering.
II. Do I bounce with my mix buss FX chain or bypass the chain?
I have seen experienced engineers use both approaches. That is to say, bounce stems with a mix buss compressor like an SSL for example. And I have seen engineer bypass everything on the mix buss.
I think the approach depends on whether you are going to master it yourself or have someone else do the mastering.
If you are going to master it yourself, and you love your mix buss effect, then I could see leaving the mix buss FX on while bouncing stems -- EXCEPT FOR THE LIMITER.
The limiter should always be bypassed when bouncing stems for mastering. Otherwise, you will have petrified stems that are hard to process and have no headroom. It's pretty hard to "un-limit" something.
So if you are the one who is doing the mastering, bouncing the stems with the mix buss chain (minus the limiter) can give you a "head start" and could potentially be a time-saver.
However, if you are sending the stems to another mastering engineer, then I recommend that you bypass everything on the mix buss chain. Yes, everything. By bypassing your mix buss chain, you give the mastering engineer much more latitude to process the stems to his/her taste.
III. Which Stems Should I make? How many is too many?
The answer to this question is personal and contextual, but here's my take on the subject.
When I started using stems back in the day, the trend started with an instrumental and an accapella. That was it. You could re-balance the vocal level, so some processing, but you were married to the relationships in the instrumental.
Then we started breaking mixes into 4 stems: drums, bass, instruments, and vocals. This trend provided a lot more flexibility for re-balancing and processing and the trend remained in place for quite a while.
The first permutation to this trend started when mastering engineers wanted control of the kick by itself, and began to request that the drum stem be submitted in two parts: kick alone, and all drums minus the kick.
Having control over the kick made the mastering engineers super happy. They could finally master the track they way they were hearing it and could compensate for limiter squeeze.
From there, the floodgates opened up. And the trend moved toward lots of stems, including separate stems for each instrument.
For me, I actually like have all of the separate stems. I know it takes are really long time and rendering stems in this way is painstaking and tedious.
But the time spent here can save you time down the road. If the mastering engineer has all of the separate stems, there is almost nothing he/she cannot achieve.
I know some mastering engineers want to keep it simple and don't want all of the extra moving parts. But I love having them. My policy is that I pretend I don't have them and proceed like a stereo mastering -- but if something sticks out, I spot fix it.
This way I don't channel strip process every stem for no apparent reason. I wait for issues to pop out and I can go directly to the tracks I need to adjust.
IV. Should the stems have reverb and delay or should those be separate stems?
Yes. Each stem should have all of its channel strip FX and reverb and delay included in the stem. Don't bounce dry stems and separate reverb return channel tracks unless you really don't like your reverb choices and want to redo them entirely in the stem mastering.
But you if like your reverb and delay work. Keep it! And bake it into the stem.
Very cool things happen in stem mastering when you apply saturation, compression, EQ and even sidechaining to stems that contain reverb. It can add density to the "air" and can give the master a very finished sound.
Furthermore, if you use a dry stem and a separate reverb return track, any tonal changes you make to the dry stem won't be reflected in the reverb track, thereby creating a big disconnect in reverbs in the master.
V. What sample rate and bit depth should I use to make the stems?
Use whatever sample rate the mix session is using. If its 44.1K, stay 44.1K. If its 96K, stay 96K.
For bit depth, goes as high as you can. 32-floating point is the best. If you have that option, use it. If not stay at least 24 bit.
VI. Do I need to dither the stems? Normalize?
No to both questions. No dither. No normalize.
VII. Should I use stereo or mono tracks?
Use stereo file format for all tracks -- even if the content is mono. A mono track like a vocal recording can very happily live on a stereo file format. It does not "hurt" the mono track and it can make it easier to use stereoizing plugins.
Juggling mono and stereo file formats gets annoying, and the DSP you theoretically save does not warrant the downsides. So in short, stick to stereo file formats for all tracks.
VIII. Bounce from zero at session start? Leave space for reverb/delay tails?
Yep. Always bounce all tracks from "zero" in the session. Do anything other than this is just asking for trouble.
And yes, leave some extra space at the end of the selection of your bounce for reverb tails and delay decays. Always better to have a little extra space than not enough!
IX. How do you preserve the side-chaining?
One of the biggest mistakes people make is not preserving their side chaining during stem creation. And that's a shame because side chain creates the groove of the music and prevents collision distortion. When the sidechaining is missing from the stems, it's really a mess.
To preserve the sidechaining make sure:
1. that all sidechain sends are PRE FADER.
2. that you check the preferences in your DAW that muting the track does not mute its prefader sends. DAW's handle this situation differently. So look into it.
Set it up. And then test it. Do some overly extreme sidechaining and then check your stems. You will hear if you preserve your sidechaining. This is absolutely critical for making excellent stems.
X. How do you deal with stems from parallel busses?
Here too you need to make sure that your DAW preferences are set not to mute pre-fader sends when muting the source tracks.
So, for example, to make a parallel drum buss stem, you mute the source faders, but the pre-fader signal is still going to the parallel buss so you can then render the parallel drum buss stem perfectly in isolation. (We show this in great detail in the newly-updated Next Level Mixing II video course).
Ok, that was sort of a long discussion, but I was trying to be comprehensive, and address all of the questions I have been receiving and give you my current best practices for stem creation for mastering.
If you have any further questions or thoughts about this topic, just hit reply to this email.
Most of all, Happy Holidays to everyone! And Happy New Year!
All the Best!