In this tutorial, I’m going to run through how to make a self-master of your audio. I’ll be using Ableton Live, but similar principles can be used in any digital audio workstation (DAW) software such as Logic, Cubase, Pro Tools, Reaper and so on, or using a standalone software mastering package such as iZotope Ozone.. The tutorial assumes you’re working in software using digital audio.
This tutorial is a long read as there’s a lot to cover. So make yourself a coffee…and then let’s dive in.
Mastering is the process of finalising audio in preparation for its intended use, e.g. being cut to vinyl or distributed by streaming services. Typically it is done by skilled mastering engineers with specialised knowledge and equipment, who charge a fee.
To be clear, I wouldn’t suggest that self-mastering is a substitute for professional mastering in the majority of cases. An experienced mastering engineer will be able to do the job to a high standard, and won’t have the inherent biases you bring to your own work.
Nevertheless, self-mastering definitely has its place in the era of digital music production. It’s good for situations where you want something close to a mastered version, but don’t require a full pro master. For example:
- If you want to try out some of your own productions in a DJ set, and need a version where the loudness is comparable to released music.
- If you’re sending out demos and you want them at a good level of loudness (I’ve seen some labels say they prefer to receive self mastered demos, and others say they don’t want any mastering on a demo, so this varies.)
- If you have audio that you want to sound good, but where pro mastering would be excessive, e.g. for uploading audio clips to social media, running a DIY podcast, releasing your own music on a personal Bandcamp page, or small projects where the budget won’t stretch to mastering.
I’ve also found that self-mastering seems to help my mixes. It gives a rough approximation of mastering, and this often reveals mix problems. I can then go back into the mix and fix it.
What is involved in making a self master?
As a general rule, it’s best to do as little as possible because it’s very easy to ruin a track with too much processing. If you find you need to make major changes to get a track sounding OK, there might be problems that would be better fixed by going back to the mix.
When it comes to choosing the right tools, there are mastering programs such as iZotope Ozone and T-RackS that can work as standalone all-in-one solutions or as a suite of plugins covering all the required functions. Ozone has a master assistant takes you through the process and simplifies it, and is a very popular choice for self-mastering. iZotope also sell a low cost, reduced-features, plugin only version called Ozone Elements, which is worth considering if you want a quick, easy and inexpensive way to make self-masters. If the process outlined in this article seems too long and complex for you, I’d suggest you get a copy of Ozone and give the mastering assistant a try.
I do professional digital mastering, so I have a range of specialised plugins for that, so these days I tend to self-master using my own customised mastering chain. But in the past I’ve done it using a more basic setup, e.g. using built in plugins in Logic, and managed to get passable results. The basic principles are similar, whether you’re working with Ozone, T-RackS or separate plugins.
The process I recommend is:
- 1. Make a pre-master of your mix
- 2. Listening, referencing, metering
- 3. Equalisation (EQ) – making small adjustments
- 4. Limiting – to get your track to the required loudness
- 5. Output the self-master – in the formats you require
There is also a school of thought that says you should start with limiting first and then add in EQ and any other processing before it, because you need to hear how any EQ changes will affect the limiting. You might like to try this. For the sake of simplicity, in this tutorial I’ll stick with the order listed above.
Many other things that can be added into mastering, such as compression, saturation, mid-side processing, stereo widening and clipping. In this tutorial, I want to focus on the bare minimum required to do a basic self-master, so I won’t discuss these optional extras.
Let’s go through the steps in order. I’ll be working on a short section of techno track of mine called Quantize. This track was released digitally with professional mastering, so we’ll be able to compare the self-master with the pro master at the end.
(The audio clips in this tutorial are 320 kbps MP3s, so they won’t give a completely accurate representation of the full quality uncompressed audio of the self-master, but should be close enough to give a sense of what’s going on.)
1. Make a pre-master
You can set up self mastering on the master bus of the DAW project where you’ve created your mix, or you can set up a separate project to do the mastering. The first option means you have the mix available to fix issues, whereas the latter has the advantage of creating a bit of ‘distance’ between the mixing and mastering stages.
For the purposes of this tutorial, we’ll use the second approach, mastering separately to mixing. So we’ll begin by bouncing down the mix as a 24 bit wave or AIFF file, aiming for it to be peaking around -6dB. A little higher is fine, but no more than -3dB. The reason for this low level is to avoid any kind of distortion, which in some cases can happen at levels below 0 dB.
The resulting audio is called a pre-master, i.e. a finished mixdown ready for mastering.
In general it’s best to make the premaster at the sample rate at which you’ve recorded the track. For me that’s 44.1kHz. It’s best to use 24 bit for a pre-master, as this allows for greater dynamic range than a 16 bit file.
Usually a pre-master should have no processing on the master bus, with all processing left to the mastering stage. Like all rules, this can be broken, but for this tutorial we’ll keep things simple and bounce the pre-master with no processing.
The pre-master can then be dragged and dropped into a new project in your DAW. Make sure the channel you’re using is set to stereo, that the sample rate of the project matches the sample rate of your pre-master, and that there are no additional options enabled that could mess up the audio.
As I’m using Ableton Live, I’ve double clicked on the audio file to open up the additional settings, and turned off the ‘warp’ mode which changes the tempo of the audio – not what we want for mastering. I’ve also made sure high quality playback (HiQ) is turned on.
Here’s a clip of how the pre-master sounds at this point. You may notice it sounds quiet, which is because it’s peaking at -6dB to give me plenty of headroom for mastering.
2. Listening, referencing, metering
Before making adjustments, listen carefully to your track, paying close attention to the whole frequency spectrum. Switch between monitors and headphones if you can, change the monitoring level from loud to quiet and in between, and shift the focus of your attention repeatedly between the low end, the mids and the high end, and then the whole track altogether. As well as listening for what frequencies might be lacking, try listening for what there is too much of, for what sticks out or disturbs the mix.
Listening to my premaster, the overall frequency balance of the mix sounds quite good to my ears. The kick thumps but doesn’t dominate, which is about the right balance for this type of atmospheric techno. In the mids and highs, the mix is a bit polite or slightly muffled, so might benefit from a slight lift in this area. Listen to the pre-master yourself and see what you think.
The next step is to do some reference listening. This is crucial. As you listen to your own track repeatedly adjusting the mastering settings, you can lose perspective. A reference helps re-calibrate your listening.
Select a few released tracks that you think are good examples of the sort of thing you’re aiming for. For my self-master, I’ve chosen one of my favourite recent ambient techno tracks, and downloaded the WAV file so I have it at full quality.
Released tracks will be mastered and therefore will likely be much louder than your pre-master. To compare them for the purposes of EQing, you’ll need to reduce the playback volume of the reference track as compared with the pre-master, until they sound at a similar level of loudness.
There are plugins you can use to help A-B compare your track with references, such as MetricAB. (I find this to be excellent, and while the full price is expensive, Plugin Alliance have regular sales and big discounts that make it much more affordable.) A simpler method is to simply drag the reference into a separate track in the DAW, adjust the level so it sounds approximately as loud as the pre-master, and use mute / solo to compare.
In my case, comparing my track with the reference I’ve chosen reinforces my initial impression that my track might benefit from a slight boost in the high mids and maybe also a little high end sparkle.
At this point, it can be helpful to add a frequency spectrum analyser to the end of the mastering chain. Visual metering is no substitute for careful listening, but can help confirm or contradict what you’re hearing. Sometimes a spectrum analyser can reveal problems that are inaudible, such as very low or very high frequencies, or resonant peaks that might need to be tamed. You can also run your reference track through a spectrum analyser and see how it compares to your track.
Most DAWs will have some kind of frequency analyser. In Ableton Live, you can find a spectrum analyser plugin in audio effects > utility > spectrum. In Logic there is a Multimeter plugin which does frequency spectrum analysis and various other useful things like loudness metering and checking stereo phase coherence. Another good option is the SIR Audio Tools spectrum analyser, which has a free version.
Here’s a snapshot of how my track looks in Ableton’s spectrum analyser:
The big peak around 50Hz is the kick drum. The area from around 2kHz to 10kHz seems slightly recessed – I wouldn’t want to massively boost this area as it could make the track sound brittle or tinny, but it confirms my sense that a gentle lift here might even things out. There also are a few strong peaks around the 1kHz area, which might benefit from a little cut – or at least, I’d want to avoid boosting that area too much.
Another important area to look at is the very low end, below 20-30Hz. These frequencies are infrasonic, so will be mostly inaudible, but they use up valuable headroom, so it’s common in mastering to use a high pass filter to reduce them. In this track, you can see that the level around 20-30Hz (at the far left in the snapshot above) is higher than the level in the high mids and highs (2-10kHz, towards the right), so this low sub is definitely something we should try to reduce.
The same principle applies to frequencies above 16kHz or so, towards the top of the human hearing range. If the spectrum analyser shows a lot of energy in this area, then some filtering here might be worth considering. In my track, the frequency plot drops off upwards of around 10kHz (the far right of the screenshot above), so there’s nothing that needs fixing here.
We’re now ready to add some equalisation (EQ) into the master chain.
Everything we’ll do here is informed by the analysis from the previous stage. That’s crucial – without a diagnosis of what the track needs, EQ is a bit like shooting in the dark, and can make things worse.
First, I’ve added Ableton’s utility plugin to the channel, and use that to make the bass mono. Human hearing can’t process directionality at low frequencies, so there’s little sense in having any stereo width down there. The utility plugin has a handy ‘bass mono’ feature and the default setting of 120Hz (i.e. everything under 120Hz is made mono) sounds fine.
Next I’m going to try using Ableton’s stock EQ8 plugin to address the frequency issues identified in the previous section. That might give a pro mastering engineer heart palpitations, because EQ8 isn’t really designed for mastering! But the point of this tutorial is to keep things as simple as possible, with the aim of creating a rough self-master that is ‘good enough’.
To my ears, EQ8 sounds excellent. It’s neutral, clean and transparent, which is helpful for mastering. It also has the flexibility to set each band as a filter or peaking / bell curve EQ, with precise control over the resonance or Q of the peaks (i.e. whether they are broad or tight).
If you’d prefer to use an EQ plugin that is more specifically tailored for mastering, there are lots of third party options. iZotope Ozone includes a versatile EQ module. FabFilter’s Pro-Q 3 is another popular choice. (Both Izotope and FabFilter offer substantial education discounts for students and staff, if that applies to you.) If you want an inexpensive alternative to Pro-Q, SIR Audio’s StandardEQ might be ideal. All of these have spectrum analysers built in. One of my favourite mastering EQs is the Brainworx bx_digital (available via Plugin Alliance and on UAD), but it’s quite complex to get to grips with. There are debates about the merits of linear phase versus minimum phase EQs for mastering, but for the sake of simplicity I won’t get into that here.
Whichever plugins you use, some useful general principles for EQing when self-mastering are:
- Do as little as possible. If you need to make big adjustments, it probably means the mix needs some work.
- Aim to cut or boost no more than 2dB. For mastering, it’s typical to make 0.5dB or 1dB changes.
- Cuts tend to sound more natural than boosts. Boosting helps to add colour.
- Make the curve of boosts wide (i.e. low Q value) for a subtle effect. Cuts can be a bit narrower if you need to target specific frequencies.
- Constantly A-B compare by switching the whole EQ in and out, the individual bands in and out, and by going back to compare with your reference tracks. To compare effectively you need to try to match the levels with and without EQ. For example, if your EQ is increasing the overall level of the track by 1dB, then you should reduce the master gain on the EQ by the same amount.
After adding EQ8 into the track after the utility plugin, and with a little careful experimentation and listening to the results, here’s what I ended up with:
- A 12dB / octave high pass filter set to 25Hz to remove the very low infrasound (the slope at the left hand side). The idea here is to cut inaudible frequencies to free up headroom, so I adjusted the frequency until there was no noticeable audible change, listening on headphones with good low end extension. The difference should ideally be inaudible but show up in the spectrum analyser when you switch the EQ band in and out.
- A peaking EQ band giving a broad boost in the upper mids, centred around 4kHz. I dialled the Q down from its default setting to make this a gentle, wide boost. With a bit of experimentation, 1dB gave a nice lift without starting to sound artificial or hyped.
- Another peaking band giving a small boost at the high end, this time with just half a dB of gain and an even wider Q. This helped to give the ride cymbals a little extra sparkle but without becoming harsh. The two wide boosts joined together to make a long gentle curve rising from around 1kHz.
I also used CTRL-click (right click on a PC) to open the context menu and enable oversampling, which can help make EQ8 sound slightly smoother in the high end.
Checking the overall output levels with the EQ in and then bypassed, the difference was negligible, so there was no need to adjust the master gain in EQ8 to level match.
Let’s listen to how this sounds. Here’s the original premaster again:
And now the EQ’d version:
The differences are very subtle. If you listen from around 30 seconds, you should be able to hear how the ride cymbals sparkle a little more in the EQ’d version.
Now would be a good time to save the file, if you’ve not done so already.
At this point, mastering engineers would often add other processes such as compression and possibly some saturation, but to keep things as simple as possible we’ll proceed straight to the heavy lifting: limiting and loudness.
A fast peak limiter will enable us to increase the loudness of our audio, i.e. the average level. It does this by detecting peaks that cross a certain threshold, drastically reducing their level, and then rapidly releasing the limiting action. By squashing the transient peaks down, the overall level can be raised up.
Limiting keeps the levels strictly below the threshold set, so we can get the peaks close to zero dB (i.e. the maximum possible level in a digital system) without worrying that they will shoot over and cause distortion.
Limiting does have some downsides. By definition, it decreases dynamic range, i.e. the difference between louder and quieter parts of a track. Dynamics are a large part of what creates variation and interest. Limiting also tends to change the tonal balance of a track. Used excessively, it can make a track sound crushed, flat or tiring to listen to. There’s a good article about this here from Bob Weston’s Chicago Mastering Service.
The key point is that there is always a trade off between loudness and dynamic range. More of one will mean you have less of the other. So for limiting our self-master, the aim is to get a good balance between loudness and dynamic range.
The stock limiter in Ableton Live is a bit basic for mastering, so I suggest using a third party plugin. Again, T-RackS and Ozone are popular choices, and both can work as plugins as well as being standalone software. Ozone has a maximiser module designed for limiting (it’s worth knowing that the Standard and Advanced versions have some improved, more transparent algorithms in this module, as compared to the more basic Elements Maximiser). I’m going to use FabFilter Pro-L 2, which is another high quality and widely used mastering limiter. It has a good balance between ease of use and advanced features.
Other mastering limiters include Tokyo Dawn Labs Limiter 6 GE, DMG Limitless, Eventide Elevate, Weiss MM-1, the Oxford Limiter and the BX True Peak. Any of these should do a decent job, so just go with what you have available, or try some demos and see which you prefer. The Waves L2 and L3 limiters are classics, but tend not to be as transparent as some of the newer options. If you’re looking for a no cost option, an earlier version of Limiter 6 is available for free.
You will also need a loudness meter. Pro-L 2 has one built in, but if you’re using a limiter that doesn’t offer this feature, you should add a separate loudness meter plugin after the limiter, such as the one that comes with Logic Pro X, or the Youlean loudness meter (a free version is available).
The next stage is to pick a loudness target. This will depend on the type of material and the intended use. Youlean has a helpful table listing loudness standards for various platforms and streaming services. Loudness is measured in Loudness Units Full Scale or LUFS. As a rough guide:
- A common loudness standard for streaming music is -14 LUFS with a true peak of -1dB. This is used by YouTube and Spotify, and is not far off the equivalent standards for Apple Music, Deezer and so on. Many of these services now have level control mechanisms that will adjust your music to their preferred loudness, so it makes sense to aim for -14 LUFS if streaming is your primary aim for your self-master.
- For dance music, if you want to include your track in DJ mixes or send it out to DJs, it’s typical for loudness in digital masters to be higher, in the range of -12 to -10 LUFS. To my ears, -10 LUFS can start to sound crushed, so I would suggest using that as a maximum, unless you deliberately want to make your self-master hammer at the expense of dynamic range.
- If you’re mastering spoken word such as a podcast or video presentation, a good loudness level is somewhere between -20 and -16 LUFS with a maximum peak of -1dB (this is the AES streaming standard).
Let’s go back to my reference track and run that through Pro-L 2 with no limiting engaged to see what the loudness is, for comparison. You could try this using a loudness meter on a few tracks in the style you’re aiming for and make a note of the numbers, to get a sense of what is typical. As you can see in the screenshot below from the bottom right of Pro-L 2, I’ve set the meter scale to ‘loudness’ and the meter time scale to ‘integrated’, which will give us an overall loudness measure for the whole track. Having played the reference all the way through, it’s coming out as -9.8 LUFS. For the sake of simplicity in setting a target, I would round that down to -10.
So let’s now go back to my self-master clip, and aim for a LUFS of around -10. It’s worth bearing in mind that the reference has been mastered by a pro, so with self-mastering you might struggle to get to -10 LUFS without audible artifacts. It’s just useful as a guideline. I suggest starting by setting up the peak limiter as follows:
- Set the metering to loudness, integrated.
- Turn true peak metering on and true peak limiting on. This means the limiter will be very strict in ensuring the audio doesn’t go beyond the maximum level set at the output.
- Turn oversampling on. This helps the limiter to catch the peaks. High levels of oversampling can introduce softness into the sound, but the 4x option in FabFilter seems acoustically transparent to me.
- Turn lookahead up to around 4ms. Again this makes it easier for the limiter to catch the peaks.
- Adjust the output level just below the meter to -1dB. This will give us a -1dB true peak level, which is a safe maximum.
- I suggest turning dithering to 16 bit, in preparation for the final output stage. Dithering is a complex topic, too much to get into here – this setting should help avoid problems.
If you’re using a different limiter, just approximate these settings as best as you can.
With the limiter set up, we can now get to work:
- Set the track playing and raise the input gain of the limiter (in Pro-L 2 this is on the left hand side of the plugin window). Bring it up gradually until the meter starts to read somewhere close to the target LUFS.
- Then adjust the limiting style and other parameters while listening very carefully to how the sound changes. For example, I often like the results of the ‘transparent’ style in Pro-L 2, but on this track, with that style selected, as I raised the gain the kicks started audibly distorting. Switching through the other styles, ‘dynamic’ gave a more transparent result.
- I generally find that the attack, release and channel linking controls work well left at their defaults, and that was the case here, but you could try adjusting the attack and release to hear how it affects the sound.
Here’s how the final settings looked:
I found that at -10LUFS the track was starting to sound sightly crushed with the dynamics squashed, so I backed off the gain a little until I was happier. That gave an integrated LUFS of -10.9, which seemed close enough to the original target.
Limiting can often reveal mix problems. Or it might reveal that you’ve been over zealous with EQ in the previous stage of the mastering chain. If that’s the case, you might choose to go back to make some adjustments at this point. Problems are always better solved in the mix than in the master.
Once you’re happy with how the self-master sounds, save the project again then proceed to the final output stage.
5. Output the self-master
I recommend outputting self-masters as WAV files. These are large but high quality because no data compression is involved. AIFF is another option, but I know DJs who’ve had compatibility problems with them on some digital DJ equipment. Most pro mastering engineers seem to use WAVs.
You might also want to output a compressed version at this point e.g. as an MP3 or AAC file, depending on your requirements.
In Ableton Live, select the audio of your track in the arrangement window (this will ensure the whole track is output – no more or less). Then go to the file menu and select Export Audio / Video, and adjust the settings as needed.
For the purposes of this tutorial, I’ve opted to export as a 16 bit 44.1kHz WAVE file. There’s a lot of debate on this point, and I don’t want to get into that here. Having got the dynamic range where we want it with the limiting, my view is that there’s no need for 24 bit, 16 will sound fine, will be more compatible with different devices, and will reduce the file size without any audible loss of quality. I’ve selected no dithering in the export settings, because I set up FabFilter Pro-L 2 to perform this function. I’ve also chosen to output a 320kbps MP3 so I have a version that is more suitable for to uploading to this web page.
So – drum roll please – here is the final self-mastered audio clip (MP3 version):
And here is the released pro mastered version for comparison:
I’m happy with this result. I actually think it’s an improvement on the pro master, which has audible distortion on the kicks (maybe from limiting or the use of a clipper). This shows that with some careful listening, attention to detail and a few well-chosen tools, it’s possible to produce a self-master of sufficient quality for the purposes outlined at the start of this tutorial.
Thanks for reading this tutorial. If you have any comments, questions or suggestions of your own tips for self-mastering, please feel free to leave them below. If you’re interested in learning more about audio production, I also offer one to one tutorials. Full details here.