When making a mastering chain, first understand the routing of your DAW, then choose a first insert that will set you up well for further processing. Typically your mastering chain will include subtractive EQ, compression, saturation, low-level enhancers, and a limiter, before including automation to make the master unique.
Before we begin to pick out our processors, it’s important to understand what we’re trying to accomplish and the general routing. Each processor we include is going to affect our signal one at a time and in the order that they’re inserted - all busses will occur after our last processor.
I like to start my processing on the channel, then include busses if needed, before finishing my processing on the master output. Let’s listen to our mix, and then master so that we understand where our session is heading.
Picking the first insert in a mastering chain can be challenging since it’ll determine the direction of your master and affect the rest of the chain. A good first insert is either a subtractive EQ, one that attenuates parts of the response, or a mastering preamp.
I prefer subtractive EQ with which I’ll attenuate parts of the signal that I don’t want to amplify. This sets me up well for having a successful signal chain later on.
Let’s listen to this EQ, and remember the effect will be subtle since we're mastering and it’s the first insert.
Sometimes a mixing engineer won’t account for how abrasive sibilance will become when the overall level of the mix is increasing - this means we may need to do some de-essing. If needed, I like to add my de-esser after the subtractive EQ, to dynamically attenuate any harsh frequencies.
You could use your EQ to attenuate this, but then that frequency will be attenuated at all times, making a de-esser a better option. Let’s listen to the effect this de-esser has on the mix.
Both compression and saturation will control dynamics from the peaks down - compression is more flexible, letting you compress from the average loudness, change the behavior, and so on, but saturation is more program-dependent and sounds a little more natural while adding harmonics in to fill the sound.
I typically don’t compress a full mix and instead use saturation, but you can use both to first control dynamics is a very accurate and specific way, then add some harmonics and analog sounding dynamic control.
When creating a mastering chain, the saturation type you use can greatly affect the timbre of your master - tape and transformers compress moderately and can attenuate higher frequencies, resulting in a mellow sound. Tubes compress slightly, accent transients, and emphasize both the low end and high-frequency range.
I find I use tube saturation most often, and lean toward warm tube saturation to include a second ordered harmonic, but use your ears and carefully decide on which one works best for the song.
Let’s listen to some subtle tube saturation on this mix.
You may be curious if you should add reverb to your chain, and in my honest opinion, it shouldn’t be added. Typically it’ll disturb the spatial relations that have been established when mixing, making for a washed-out sound, or one in which the dimensionality of the mix is less apparent.
That said, use your ears and decide what’s best for your session. Let’s listen to this chain with subtle reverb added, and know that I’ll remove this for future demonstrations in this video.
We typically think about controlling dynamics from the peaks down, but there are some ways to increase the signal from the quieter details upward. I could use an upward compressor to measure, compress, and amplify quieter parts of the signal, but instead, I’ll use the Oxford Inflator.
This processor prioritizes bits of information with a positive value over those with a negative value, meaning it’ll increase the ratio of signal to silence or high-level signal to low-level signal.
In my opinion, and according to the developers, the best way to use this processor is to put the effect to 100%, and then adjust the input to determine how aggressive the processing is.
At this point in the chain, it’s a great idea to include another EQ with which we’ll amplify parts of the signal we want more of. Since our ears are much more sensitive to 3 to 5kHz, we could amplify this to make the mix sound a fair amount louder.
We could also amplify parts that are hard to hear, like the super-high frequency range to add air. Furthermore, we could use mid-side EQ to narrow or widen the stereo image by amplifying the mid or side respectively.
I like to amplify some of the side image’s highs and mid frequencies, then amplify 3-5kHz on the mid, and maybe some of the kick’s fundamental.
Let’s listen to the effect this additive EQ has.
The limiter you choose is going to have a huge impact on your sound - I personally enjoy limiters with multiple algorithms so that I can cycle through and find one that works best. My recent go-to has been the Oxford Limiter, with its enhance feature at a higher level.
My favorite free option that I’ve found is the CIS DSP Factory LimiterOne, which has a really aggressive and detailed sound.
Let’s listen to the Oxford limiter and use about 3-5dB of attenuation, and notice how the signal is certainly louder, but also, more impressive regardless of the additional loudness.
Now we have everything in place, and it’s time to make the master unique by having the processing respond to the emotionality and intent of the song. To do this, let’s automate in and out various features to increase or decrease the loudness, fullness, and other parameters.
I think a great way to make the master more engaging is to increase saturation or upward processing during choruses or maybe just the last chorus to differentiate it.
Let’s listen to both the saturation and the enhance feature of our limiter being increased to better understand the purpose of automating our processing.
Last up, let’s consider what metrics we should keep an eye on - I, like most engineers, will use an LUFS meter to understand the perceived loudness. For this master, I’ll make it -14 LUFS, since YouTube is going to try to normalize this video’s entire level, not just the demonstrations.
When it comes to peaking, I’m really not too concerned with this unless it’s happening on the master output. Due to 32-bit and 64-bit internal processing, our signal can go above 0dB without clipping, so long as it’s returned to below that level before the master output.
Let’s take one more listen to our before and after of our master, and keep things like the frequency response, the dynamic range, and the level of distortion in mind.