Mastering should be left to professionals, with the right training, skills, and tools.
That being said, for a lot of my (and I guess I can say “our”) projects, spending the money that a proper mastering job would require is not available. In part, that has also to do with part of my way to do things: if I declare a project as “no-budget”, then it does not have a budget (with the exception of donations, which mostly come in the form of workmanship by invididuals, and for my solo efforts, that individual is mainly me).
So what can we use if we have to cut corners financially, and want to do the job ourselves?
A lot has obviously changed to the better with the advent of computer plugin technology, be they VST or some other format. While in the analogue past – and also in the analogue present – a single component could easily set you back a few thousand bucks, today you may very well decide to do it all in the computer, and as that keeps you from paying for expensive electronic components, you can easily get the plugin equivalent of that EQ monster for much less than one tenth of the price.
What’s more, you can get another fancy plugin for one tenth of that price, and another plugin for the same role for even less, namely for free.
So what’s there for your mastering pleasures?
Where this is coming from – a.k.a “Scope”
I’m currently working on the master for the oscillator theory album. I’m doing this myself, I’m doing this in a software-based environment, and I wanted to see what kind of tools I could use.
There’s three things you need to know about my setup:
- I’m doing this using a dated version of WaveLab, namely version 6. So everything already included here is considered “free” in the context of this article, although that only makes sense for someone who already has this (or another) tool.
- My mastering approach is very minimalistic. There’s three reasons for that.
- What goes into mastering is in a proper shape. If there’s an issue in the mix, I don’t fix it in mastering.
- I believe in a transparent mastering job. Most of the time in mastering is spent with track gaps, fades and track levels. I don’t want it loud. K-14 usually is plenty.
- Your mastering chain is complete when there’s nothing left to remove. As few tools as possible. In an ideal project, I’d record a great acoustic ensemble in a great room with a pair of nice spaced omnis thru a proper pre, and release that.
- Source material is typically 32bit float, 48kHz, stereo. Masters are 24bit, 48kHz, stereo (unless it’s for CD or MP3, in which case that becomes 16bit, 44.1kHz, stereo). And I don’t change that along the chain.
With that being said:
The Generic Mastering Chain
The figure above shows what I call the generic mastering chain.
The “Source” and “Master” blocks are essentially the files the audio is going to and from.
“Levels” is the level of the individual track – which I typically set in the DAW. So this is not a plugin job.
The remaining five blocks are typical plugin chores – three of which (the ones with the dotted lines) are considered optional. Note that in this chain, the “Dynamic Colour” and “EQ” block can also swap position, depending on the job at hand – the other blocks are very much fixed in their position.
“Dynamic Colour” is what you typically do with some kind of compressor with (for a mastering job) very short time constants. It can well include driven tubes, tape saturation, and transient shaping. I mostly either don’t want those, or if I do, I want them in a very selective fashion – that’s why it’s done in the mix – and here, marked as “optional”.
“EQ” should be clear: anything you do to the frequency response. Note that in some cases, this may also include nonlinear processing (e.g. EQ saturation).
“Limiting” is to make sure that nothing hits the digital redline. The thing gets louder on the way, which often is a very desirable side effect. Usually you won’t go without it. Note that the limiter is in the right place after the sample rate conversion, to deal with intersample peaks and all.
“SR Conv.” stands for sample rate conversion, and is just that. Which also means you don’t need it if e.g. you’re going form 48kHz soure to 48kHz master.
Finally “Dither” may also include noiseshaping. While the general agreement is that dither shall be applied for 16bit masters, it’s debatable if that is also true for 24bit masters. For that reason, it’s optional (although I tend to use it, and I always use it for 16bit targets). That step is always the last in the chain.
All in all, we have six processing steps, one of which is not a plugin job, and three of them are optional. Plus, for two of those (SR Conv. and Dither), I have nice tools coming with my DAW. So the items of interest are EQ and Limiter (for both of which the onboard tools of Wavelab 6 are rather…unsatisfying).
Coming from the collaboration of Variety of Sound and Tokyo Dawn Labs aka vladg/sound, SlickEQ is a three-band semiparametric EQ plus highpass with a few tricks up its sleeve.
There’s three overlapping bands with adjustable gain and frequency, and a frequency range per band of almost two decades, plus a totally un-masteringy gain range of +/- 18dB. High and Low bands can be switched from bell-shape to shelving mode, and finally, there’s a third-order highpass. No fancy graph displays, just use your ears.
Some of the beauties are behind those EQ bands, however. There’s a distortion output stage, offering different characteristics (called “Silky, Mellow, Deep”), adjustable drive level, and the option to turn that off. The EQ bands can also be driven if set for positive gain. There’s an option to only affect L, R, M or S channels, allowing for individual processing if more of those are chained.
The nicest option is the choice of EQ models, aptly named “American”, “British”, “German” and “Sovjet”, all of which give you different EQ curves.
One word of caution, though: the drive and saturation models don’t just distort an input signal, they add some noise of their own. Which is a bad idea with if you have a dither thing with autoblack coming afterwards.
And finally, it all goes out in 32bit float. And it’s free.
Verdict: if you are not the analytical type, if you know that two shelves and a wide-range bell are typically enough for mastering jobs, and if you might want some sonic coloration from time to time, this can be your option.
Recommended by the folks in about every knowledgeable online forum, this is the exact opposite of the one mentioned above.
Coming from DDMF, who is always moving between super-analytical and totally crazy plugins for EQ and compression duties, this is a very analytical thingie:
Ten bands of EQ, each of which can be set to one of ten EQ types. Choices include standard architectures and Butteworth designs with variable slope up to 60dB/oct. Phase can be set to minimum or what’s called “linear” (street vernacular for “constant group transit delay”). Drag-around UI availble, as well as a curve of the EQ, and there’s also a spectrum analyzer.
While SlickEQ is more one of that studio mojo things, this one is the audio counterpart of a six-axis CNC milling cutter. Adjust the finest details. See what you’re doing.
Downside? It looks more like a lab than an art space. And it costs 39 bucks – unless you wait for the next black friday/christmas/end of year sale or so, in which case it’s typically half of that. Great value, though.
Yet another gem by vladg/sound, this limiter (here shown: the more simple GUI #2), which also comes with a lot of net cred, is actually more than a limiter, namely a program compressor, two limiters and two clippers.
Leaving the compressor (a fun component in its own right) aside for the moment, we first have a limiter with different Mode/Type settings to choose between various timing characteristics – giving you the option to avoid perfect brickwalling in this stage, which can be sonically advantageous. An additional HF limiter allows you to tame those very aggressive HF peaks (which, incidentally, are usually the result of a bad mastering job). The first clipper allows for 4x oversampling, and to be run in multiband mode. Finally, the ISP module is a true peak value limiter which will catch those peaks happening in between samples, especially if the first clipper is run in oversampling mode.
All in all, what I like about this thing is that it does what I want a limiter for: it’s very transparent, and if you don’t decide to hit it too hard, you’ll hardly hear it working. And it’s a 32bit float external interface again. And free again.
Reason should have it that I never use a tool with a name like this. But I still like it.
User interface is deceptively simple: two sliders, one for threshold, one for ceiling. The “Link” button allows you to gang those two. Three displays, for input, output and gain reduction. Don’t know why two are horizontal and one is vertical. This is like a Waves L2 without the release control and IDR options.
The nicest thing about this one seems to be that it has some kind of intrinsic loudness war protection in its algorithm – meaning that even if you really turn it up, it will at one point simply stop to further increase loudness, rather than introduce wholesale distortion.
In difference to the Limiter No6, which has so many options and functions to choose from, this is the simple version: Set ceiling. Set threshold until you arrive at the desired loudness. Done. And, it’s also free. Nice!