Divide and Conquer

Sep 1, 2004 12:00 PM, By Julian McBrowne

In the not-so-old days of not-so-total recall, engineers hedged their bets by printing a slew of alternate mixes at the end of every mix session. They'd print mixes with vocal up, vocal down, bass up, bass down, and so forth. If a revision was called for, and you were lucky, you had an appropriate version tucked away on a DAT. If not, it was back to the studio for an expensive recall session. Somewhere I have a listing of 21 alternate mixes that one record company required for a final-mix delivery.

If you're mixing from a DAW multitrack, in which syncing up multiple files without adding additional noise is a simple proposition, a technique called stem mixing gives you the ability to generate virtually limitless alternate mixes without having to go back to the multitrack master and recall your mix session. In a nutshell, it entails separating various mix elements into subgroups to be recombined later (see Fig. 1) and printing them as separate files, called “stems.”

THE REASON WHY

Splitting your mix up this way gives you the flexibility to adjust the levels of the subgroups after the fact, making it possible to deal with most after-the-mix change requests. Stem mixing is the standard in film — in which music is often delivered as separated rhythm, instrument, and vocal submixes to go along with FX and dialog — but the concept works well for music mixes, too.

Music mixing is a delicate blend of art and science, with the goal of producing an accurate and emotionally satisfying final product. Often, at the end of a long mix session, when I'm tired, my ears are fried, and my critical faculties are reeling from paying so much attention to one thing for so long, I'm less likely to be sure about all of my decisions.

Everyone in the control room may agree that the mix sounds great, but it's smart to defer final judgment to a time when your ears are fresh and you've had a chance to listen back on several different systems. Maybe I'm picky, but I've never heard one of my own mixes that couldn't be improved in some way. Sure, great mastering can and has saved many a questionable mix, but wouldn't it be great to get one more shot at achieving that perfect balance before your mix hits the mastering room?

Read on, and I'll show you how to render your entire mix as a group of stereo stems that can be rebalanced with no loss of fidelity. These premixed subgroups of instruments or vocals will have the identical effects, level changes, and equalization settings that they had in the stereo mix. Once the subgroups are put into a new session, you can execute what's essentially a mix-after-the-mix, rebalancing and fine-tuning the blend of the subgroups. Although that does add one more mixing step before mastering, it also gives you a high level of control over the final product.

MIX AND MOLD

The first step in the process is easy: mix as usual. Put on your mix hat and go to work with plug-ins and outboard gear. Do all the balancing, muting, equalizing, and automating that you normally do. Tweak until you're satisfied, and then tweak some more. If your modus operandi includes strapping a premastering level maximizer like Waves L1 or L2, or an EQ such as the TC MasterX across the master mix bus, do it. But be prepared to document and save those settings, because all master-bus effects must be removed before you print your stems.

Once you're happy with your mix, designate a mix start and end time. Doing that now will make it much easier to line up your stems later. Pro Tools, Logic Pro, Nuendo, Digital Performer, and other sequencers all include marker functions for that purpose. In Pro Tools, I like to drop a marker at the beginning and the end of the mix, select the entire region, and create a Memory Location that I name “Whole Song” (see Fig. 2). Every time I choose this location, it selects exactly the same region with the same start and end time. Stems made this way line up perfectly every time.

Next, print a reference mix that's identical to the final mix except without any master-bus effects. Make sure to save or document the settings for these effects, because you'll need to put them back on later. Be aware that removing the master effects may cause your overall mix level to drop. therefore, resist the urge to adjust the master fader; you'll get the gain back when you reinsert the effects in your stem session.

If you're working with 24-bit files, there's no need to dither them down for the reference mix. You can convert them to 16-bit later when you're ready to create a CD from the stem session. Print the mix to your hard drive and give it a unique name that you'll recognize when it's time to import it into your stem session. This mix will be your reference for the stems.

SPLIT ‘EM UP

The next thing you need to decide is what stems to make. Your decision might depend on which levels you're most unsure of, or it might be linked to a particular musical style. Noted mixer and three-time Grammy winner Tony Prendatt suggests always doing a bass stem for R&B and hip-hop tracks since bass levels are so critical in those genres. In his most complicated mixes, Prendatt admits to preparing as many as ten stems, including separate string and horn stems. If a mix adjustment is required, “the extra work pays off, since I can avoid going all the way back to the master to fix the mix,” Prendatt says.

In my experience, the most contentious issues in a mix are the vocal levels, so you will definitely want to create lead-vocal and background-vocal stems. From this point, you could simply create one more music stem that includes everything but the vocals, or you could separate the music further and create drum, percussion, bass, and instrument stems. Some situations might require even more divisions.

It's best to create the instrument group by muting everything that's not in the group, rather than by soloing things that are in the group (see Fig. 3). It might seem that soloing would achieve the same result, but unfortunately, all solo functions are not created equal. Some DAWs have a solo master that produces a different level from the master output; others change every soloed instrument into mono. Additionally, a soloed channel might not include the signal from the effects returns, depending on how the software is configured.

One thing to watch out for is the signal from prefader sends. If you're muting tracks to isolate the ones targeted for a particular stem, and you're hearing ghostly effects returns from tracks you thought you'd shut off, remember that the prefader effects live on after the track is muted, so you'll have to turn off those sends as well.

GIVE IT A TRY

Let's go over the steps involved in doing a stem mix on a typical song. Let's start by making a drum stem. Carefully mute everything in the mix that is not part of the drum track. Leave all of your effects returns on and leave the master fader at 0 dB. Bounce this mix in stereo to your hard drive and name it so you can find it later. Let's say your song has some additional percussion instruments that you want on a separate stem. As you did with the drums, turn off everything but the percussion, bounce it, and name the file. The same process is used with bass, lead vocals, background vocals, and whatever other stems you decided you need. Listen to your stem mixes as they go down to your hard drive; you'll know right away if you've muted the necessary tracks.

As you create your stems, keep track of which instruments are in which groups, because duplication will throw the final combination out of balance. For example, if you inadvertently assign the cabasa to both the percussion and drum stems, it will play back twice as loud in the stem mix. If your DAW allows, mute and hide or disable tracks that have already been assigned to a stem group. Consider printing out a copy of your mix page and checking off the tracks as you assign them to the various stems. Another way to avoid confusion is to put the tracks for the various stems next to each other in the DAW's Track display. That way, you don't have to select any noncontiguous tracks to be part of a particular stem.

TOGETHER AGAIN

Once you've mixed all of your stems, create a new 24-bit session on your DAW. Import all of your stems, along with the reference mix, and snap them all to 00:00:00:00. If you've used the same selection region for each stem, they should all line up. Next, create a master bus, reinsert your master effects, and reproduce (or recall) your original settings. Set all of the mix faders on this new session to 0 dB, mute the stems, and play back your reference mix. It should sound identical to your multitrack master mix, and you should hear all of your gain and EQ returned through your reconstituted master inserts.

Now mute the reference mix and unmute and play back just the stems. What you hear should sound exactly like the reference. If you can, group the stems so that you can easily switch between the reference and the combined stems. Once you're sure that the stem combination is an exact replica of the reference mix, it's time to generate some 16-bit CD-R versions to listen to around town.

TAKE A LISTEN

After leaving the mix for a day or so to regain perspective, listen carefully on a variety of speakers. Most people have several places where they are comfortable checking out their mixes. I like my car, the funky boom box that lives on top of my girlfriend's fridge, and the monitors in my personal studio.

After listening, it's time to return to your stem session and make whatever adjustments are necessary. Raise the lead vocal stem, dip the backgrounds, boost the bass — whatever it takes to fine-tune the blend. It's an opportunity to apply a post-production perspective to a mix you thought was finished. This is the appropriate time to consider adjustments such as adding brief volume boosts on the vocal stem to clarify words that are unclear or buried in the track, or creating an a cappella section that didn't exist in the original mix, and so forth. When you're satisfied with the changes, you can generate the final mixes.

Whether you store your mixes on CD or on DAT, now is the time to consider printing alternate versions. I usually make the following: main mix, TV mix (a performance track without lead vocals), instrumental, a cappella, backing vocals (only), and lead vocal (only). Once your levels are set it's easy to generate all of these mixes from your stems, assuming that you have the drums, bass, instruments, backing vocals, and lead vocals separated out. For the main mix: keep them all on. For the TV mix: mute the lead vocal stem. For the instrumental mix: mute all vocals, and so on. See the table “Stem Components and Alternate Mixes” for more details. You can even make the old vocal-up, vocal-down, bass-up, and bass-down versions if you think you'll need them.

You may wonder whether it's advisable to bring a stem mix to your mastering engineer to give him or her additional control. I posed this to mastering engineer Michael Fossenkemper from Turtletone Studio in New York City. “Having a stem mix is great if you need to make changes before you get here,” he said, “but please don't bring your stems to the actual mastering session. It makes mastering too much like mixing.”

At most, Fossenkemper will accept drum, bass, instrument, and vocal stems, but he clearly prefers to work from a well-balanced stereo mix. “It's best to give me a mix that you're really confident about, but sometimes it can be helpful to get a separate vocal stem that I can de-ess, if necessary, without affecting the rest of the track.” He says to be certain to check with your mastering engineer before you show up with stems.

THE STEM OF ALL MIXES

If you're considering an eventual 5.1 remix of your project, you might also consider making a set of stems without effects. Tony Prendatt and I recently completed a surround remix from a set of dry stems that we rebalanced and enhanced with surround effects. The original mix engineer prepared an extensive set of dry stems including keyboard, guitar, percussion, bass, and chorus. Many surround projects are being built this way since it's far easier than starting a mix from scratch in a 5.1 environment.

By adopting stem-mixing techniques, you will be able to tweak your mixes without having to recall your original mix session. Although it does add another step to an already complicated mix experience, the reward is having greater control over your final product.

---

Julian McBrowne is an engineer, a producer, and a self-styled digital audio guru who lives in southern Vermont.

Stem Components and Alternate Mixes

This table shows a typical stem configuration and the types of mixes that can be generated from these various subgroups. Level adjustments are optional.

Stems	Drums	Bass	Other instruments	Backing vocals	Lead vocals
Main mix	X	X	X	X	X
TV mix	X	X	X	X
Instrumental mix	X	X	X
A cappella mix				X	X
Lead vocal mix					X
Backing vocal mix				X