Ask five audio mastering engineers what they think of stem mastering and you’re likely to get five different answers. Some love the flexibility offered by working with stems and feel it allows them to get a better end result for their clients. Others feel that using stems comes dangerously close to mixing, and that it’s not their place to alter fundamental mix balances.
Others see it as a viable method to work around specific challenges, but not necessarily an approach that should be used by default. I want to be clear up front that I fall decidedly into the latter camp and as such, my advice comes from that perspective. Still, as a “centrist” position I believe it will be valuable to you when navigating the world of stem mastering.
In this piece you’ll learn:
A stem is a recording (or print, export, or bounce) that combines multiple individual channels into a single file, typically stereo, including any processing. Most often, the channels included in a stem will all be related (e.g. drums, or vocals, or synths).
If you’d like some further reading to help solidify the difference between stems and multitracks, or give you some guidance on how to properly prepare stems, you can check out this article.
To identify when using stems may be appropriate, let’s turn this question around and ask ourselves, “when should we avoid using stems in mastering?” Here are a few examples of situations in which mastering from stems is putting the cart before the horse:
It may sound like I’m ruling out just about any case you can think of for using stems in mastering, but let’s briefly examine why I’m making these recommendations. First, the artist and mix engineer have usually worked long and hard to craft a mix they both love. As mastering engineers, we need to be careful to respect that and not make changes that fly in the face of that hard work.
Second, creating stems which perfectly sum to the mix can be harder than you’d think. Depending on how effects are routed, what sort of bus processing is in place, and whether or not outboard hardware is used, even a few stems summed together can end up sounding significantly different than a full 2-bus mix would.
So what’s left? When is it acceptable to use stems in mastering? In the words of Bill Nye, “Consider the following.”
Under circumstances like these, it may be in the best interest of the music to work from stems, or perhaps the mixer or artist is already aware of the issues and is asking if you can address them with a stem mastering approach. In either case, it’s best to proceed with respect and care.
In the context of the previous section, my answer will likely come as no surprise: only as many as are needed, or put another way, as few as possible! In many scenarios this will likely mean two or maybe three. For example, perhaps vocals and everything else, or drums, bass, and everything else.
If a situation arises where more seem warranted, it can be helpful to think about exactly why you need each stem. If you can’t think of a really good reason, then you’re likely better off combining it with another stem. Ultimately, the maximum number you choose to accept will be a very personal decision. For me, I struggle to think of any scenario where more than seven or eight would be warranted, and that’s an extreme case that comes dangerously close to mixing for my taste.
I hesitate to spell out too many specific scenarios for stem mastering as I don’t want to give the impression that it should be a go-to solution for every issue you encounter. Rather, I’d like to reinforce that it should be a carefully considered alternative for unique situations, many of which may not share much in common.
Still, I believe it will be instructive to consider a few scenarios so that you can develop a sense of when stem mastering may be appropriate. Additionally, I will suggest a few techniques for setting up a stem mastering session in your digital audio workstation.
It can often be advantageous to treat a stem master like a two-track master in a larger session. There are a few ways to approach this. The most rudimentary way is to load the stems onto tracks which are routed to a bus or group track. This allows you to make the individual changes you need to the stems while still treating the bus with your typical two-track processing. Another method, allowed in some DAWs such as REAPER or WaveLab, is to use a nested sub-project. This can allow you to work on the stems in a dedicated multitrack environment, which is then represented as a two-track bounce in your main session.
Broadly speaking, this category probably accounts for most of the sessions on which I’ve ultimately resorted to stem mastering. Sometimes it can be to aid in noise removal with RX, while other times it can be to help with level automation, compression, or de-essing when a mixer is struggling to achieve the control they’re after in a transparent way (see below). In either case, the potential to have a negative impact on any effects associated with the vocals is there, so proceed with care!
We all know bass can be one of the hardest things to get right, often because of how hard it is to get accurate low-frequency monitoring in a small room. As such, it may be that the monitoring in a mastering room makes it easier to sort out issues in the low-end. In these scenarios, stems for the drums, bass, and everything else are usually enough to rectify issues in one source without negatively impacting the others. The potential downside here is that any bus compression is probably most strongly affected by the bass and drums.
One way around this is to route a copy of the full mix to the external sidechain input of your bus compressor, ensuring that it still reacts to the full mix even when processing a single stem. Still, things tend to be intricately connected by the time we get to this stage of a mix, so caution is the name of the game.
From time to time, you may come across a mix in which the transient energy in the drums is making it difficult to achieve the level you’d like without adding pumping or distortion to everything else. In these cases, having a separate drum stem can allow you to add a little saturation, clipping, or limiting to just the drums before using compression and limiting in your main mastering chain. This should be reserved for extreme cases though, and will likely be increasingly unnecessary in today’s loudness-normalized streaming world.
This is certainly not an exhaustive list of the possible uses for stems in mastering, but hopefully it gets you thinking about how they can be utilized in a useful way.
Using stems in audio mastering can certainly open a world of possibilities. Sometimes these are genuinely the most effective ways to get the best results for a song, but if we’re not careful there can often be unintended consequences. As with so many topics we cover here, careful listening and judicious application of the techniques discussed is critical.
I want to leave you with a brief, cautionary tale. There may be a time when someone comes to you asking you to do a full stem master from perhaps six or more stems. If this happens, it's especially important to think about why they are asking you to do this. It could be that they’ve heard of stem mastering and think it’s a trendy approach they want to try, or it could be tantamount to them saying, “I’m not really sure if the mix is finished...” In these situations, it’s always worth opening a line of communication to help establish exactly what the goals are.
So, as long as you proceed with respect and care and consider the cautions outlined above, stem mastering can help you achieve the sound you’re going for.