Avoid clipping in your final mix

So you've created an absolutely banging tune and it is time to give it a final polish making it ready for mastering. But, the little clipping indicator has been lighting up while your song plays, perhaps so much so the sound of your mix sounds rough and undesirably ugly. No good in having an amazing song if the final mix sounds like a fudged up mess! It is general practice to have a really clean final mix so everything is ready for mastering (more on mastering later). This article will help you aim for a final mix that is nice and clean so it at least avoids clipping. If you follow the provided steps you might pick up on neat habits that will prevent you from sonic catastrophes in the future.

Before getting deep into this article it is best that you make sure you're fully familiar with Renoise's DSP audio mixing system, which can be found here in the Renoise Manual. If you're unfamiliar with any of the Renoise terminology used here it is always good to go back to the tutorials to brush up on some technical knowledge of your favourite music software. This article also uses a lot of common audio production terms – so if in doubt look the word up. I've linked some of the terms so you can follow up on with further reading.

What is clipping and why avoid it?

Master Gain

Clipping is a form of distortion where the amplitude of the waveform is attempting to be so large it runs out of room and pushes against the extreme edges of available sampling space (referred to in digital audio terms as '0dB'). Because it has nowhere to go the waveform is flattened out at the peak of the amplitude in a hard line until the amplitude lowers again and resumes its normal curve (see the two waveform diagrams below). Both the hard line and the fact that there is a tight little corner bent into the shape of the waveform contribute to audible distortion. The sharper that bend and the longer that line is straight the harsher the sound of the distortion. If you'd like to know more about the technical background to clipping and distortion, start reading here.

Example of a clean waveform with plenty of headroom.

This is a healthy looking waveform. Plenty of headroom here. But look at this one:

Example of a clipped waveform.

This waveform has been clipped. It is the same sound as before but with +12dB added. You can see the edges have become squared.

This of course is sometimes a desirable sound effect, but for the case of making a very clean final mix on your song it is not desirable. Not only can clipping sound ugly, it can prevent a good result in the mastering process later on after the final mix is done. Why is this so? Mastering initially requires the final mix to have at least some 'headroom' to work with. The room to move is needed when applying further effecting to the waveform, which is commonly further boosting the amplitude. Headroom is the amount of space left between the loudest peak amplitude (say -4dB peak) to the absolute edge of the digital audio envelope at 0dB. If your mix 'peaks' at -4dB, that means you have 4dB of headroom, and probably ample space for mastering to do its job. In case you haven't figured it out by now all the loudness values in digital audio are expressed in negative dB. -4dB is quieter than 0dB. Look at the following example:

Renoise master channel showing the peak level, with at least 4dB headroom.

If your own mix has peaked at 0dB and the clipping light has turned on, then you have distortion and no headroom left for mastering. Time to rethink the plan!

Ideal master channel settings

Doing mastering for clients over the years has allowed me to see some pretty varied use of the settings on the master channel, and some songs have been downright wacky! So, I've developed settings and a method that helps prevent mix clipping and leave a usable amount of headroom. First, let's look at the Track DSP strip for the master channel (Mst):

Ideal settings for clean mixes on the Renoise master channel.

Take a look at the settings in this image. Don't worry too much if you don't know what every single thing here means, we are only examining the chosen settings for this task.

Let's go through each relevant setting: Panning is centred so that nothing is pushed unevenly out to any of the sides. Pre-fader volume is set and fixed to 0dB (very important!). The Width is not used at all because we don't want to introduce strange short delays to the mix and muck with the stereo image. The post-fader is set and fixed at 0dB (again very important!). DC Filtering is off because it's better to clean up the low end frequencies selectively in the mix. Soft Clipping is off for two reasons: one is that we don't want to hide the fact that the mix could be clipping; the other reason is we don't want any subtle alteration over the entire mix potentially colouring the peaks in an undesirable masking manner. The Auto Gain is off so that Renoise is not altering the post-fader volume behind your back, thus making any potential clipping appear straight away. Overall this approach is very transparent – letting you hear the mix as is without any alteration.

I can hear some of you say "Hold the bus! I don't set up my mix this way" Well, that's ok, but if you want to get a very clean mix without clipping this is the best way. It may now become apparent your song sounds too quiet... Well, time to turn up your monitors! This brings up a wider issue about monitoring.

How to set up your monitoring levels

You may have heard of the term 'bit-perfect', and if you would like to learn more about bit-depth issues you can delve into the subject here. Without heavily going into the details it is important to adjust your playback volume of your monitors (speaker or headphones) using an analogue fader, not a digital one. This means your soundcard channel volumes (if the card has a software mixer) all have to be set at 0dB. This will prevent bit-level degradation to the sound. Some people's studios have an amplifier or hardware mixer connected to the soundcard output before it goes the monitors – adjust the playback volume there. Similarly, if your monitors are active and have rear attenuators, you can adjust the playback volume that way. Or, you can do what I've done and get a passive analogue volume knob that sits between the soundcard output and my active monitors (see picture below of my big knob!). Whatever method you choose, it will sound better if you're adjusting the playback volume with an analogue fader – digital adjustment isn't bit-perfect and can kill representation of dynamics at settings other than 0dB. If all of this sounds a bit too daunting for you then you can just adjust the master output fader on your soundcard, and make a project of later on getting an analogue fader to do that job properly.

MMD has a big knob!!!

The other major point with ramping up your monitor playback volume is that you might use your computer system to play things other than your Renoise mixes – for example: mp3s, DVDs and CDs. These might now sound just way too loud! What to do? Because you're using the master channel settings I outlined earlier your mixes will sound loud enough, but anything that has had commercial mastering done on it is going to sound way too loud. So, you'll need to roughly decide on two monitor playback volume settings, each depending on what you're doing at the time. This is where having an analogue fader is really handy – I put mine at 11 o'clock for listening to CDs, and way up to 2-3 o'clock for working on final mixes in Renoise. Your settings will be relative to what gear you have.

Commercially mastered material has boosted loudness, sometimes excessively boosted (so much so that they've given a name to the phenomenon: The Loudness War). I want to stress that a final mix of your music in Renoise is not as loud as commercially mastered audio. Do you remember from before that a final mix needs to avoid clipping and have some headroom? This is why you will need two different settings for your overall playback volume – a louder setting for working on mixes, and a quieter one for playback of commercially mastered audio.

Final mixing – avoiding clipping

So you've got this far and set your master channel for clean uninterrupted output, and you've set your overall playback monitoring volume level suitable for working on a clean mix with headroom. How do you avoid clipping? Initially this may seem as simple as 'turning down the volume' on individual channels – but more it is more likely there are other elements you can fix up with more precision. What are some of the common things that cause mix clipping, and what can you do about it?

Excessive bass frequencies are almost certainly the number one culprit of causing mix amplitudes to be so high they are clipping. There is a whole separate art form of getting a final mix tonally right so there are not any excessive frequencies in any area. We'll look at aspects of this in future articles. Another related issue is the need for accurate monitoring – having speakers that represent all the frequencies in an honest way (which is yet another topic we'll look at later). The general point though is lower frequency sounds require more amplitude to express themselves and be heard on an equal level (see the Fletcher-Munson Curve for an explanation). Because of this need for equal expression, bass frequencies can quickly get out of hand and excessive. You may need to give attention to those frequencies by closely EQing sounds like kicks and basses. Reducing low end, especially sub-bass frequencies in sounds where they are not needed will also help clean up the mix and avoid needless excess in amplitude.

Renoise master spectrum showing a case of excessive bass frequencies.

Way too much booty!

Snappy large transients of a quick nature can also 'spike' through your mix and cause unnecessary clipping. What is a transient? Take a look at a the waveform of a snare sound – the transient is the part of the sound where it is the most 'spiked' – a quick transition of sound having a peaking percussive quality to it. Transients are very common in percussion sounds, but can also be found in snappy sounding instruments like funky guitars or plucked synth sounds. These quick sounds can sit above the rest of the mix sounds, and sometimes so much so they cause clipping. This can be remedied in a number of ways. One way is to apply gentle compression that addresses the excessive part of the transient. Compression using a shorter attack in order to be quick enough to catch the spike will make sound 'behave'. Another way is to apply very gentle and light digitally emulated saturation on the transient sound, which will reduce the snappy dynamics of the sound. It goes without saying that compression and saturation can be overdone, and that's something I'll cover in detail with future articles.

A case of excessive transients causing clipping.

There be some stray transients going on!

Summing up...

In summary: if in doubt reduce something! You may find yourself reducing just about every channel in your mix to get some reasonable headroom happening. Sound too quiet? That's ok, just turn up your monitors! The more you work according to this method, the more confident you'll get in making a slick and clean mix without clipping. You've got at least 24 bits of data resolution to play with to express all your dynamics, around -144dB worth of dynamic range! That is a lot, so use it! No point in slamming it all up near 0dB risking clipping. Some quieter styles of music don't even peak over -6dB in their final mix.

I've made up an example XRNS file that's at the final mix stage. To play this file you will need at least Renoise 2.5. It will show you an example of a simple clean mix without clipping, and you can use it as a benchmark to set your playback monitor levels. It is also an example of a neat final mix that is 'ready' for mastering. So, it won't sound 'done' because that's what the mastering achieves.

The example song also shows a little use of the master pre-fader for the purpose of a fade-out (see the last pattern). If you songs need to use a master fade-in or out, then automate the pre-fader on the master channel up to or down from 0dB. That way, for the bulk of your song it remains uninterrupted at 0dB. So make sure that your automated maximum points are precisely set at 0dB.

Example of a fade-in rising to precisely 0dB

Whether you're going to do the mastering yourself later on, or passing the song on to a mastering service – you now have a good idea how to complete a final mix without the major sonic disaster of clipping. The cleaner and better presented the final mix is, the easier and less destructive the mastering will be. Good luck! And if in doubt, share your mix with the Renoise community and you'll surely receive some great tips and feedback.

Category: 

30 comments

Great article! In my default empty track I always run a multiband compressor on 'light' settings, with a limiter set at -0.200 to avoid blowing up my ears. Didn't know Renoises Dc-filter messes around with the low freqs, will turn this option off in the future. Do you recommend having the 'dither' option on in the preferences or use a vst on the master for this?
On the final export please dither it.
On the final export please dither it.
Very straightforward and easy to understand article! Thanks for the tips on dynamics. They were solid and well articulated.
plug > I'll be covering dither in a future article. But for now I'll say that the dither setting in Renoise only applies to the offline rendering function when selecting a bit depth of anything other than 32bit float. My suggestion is to leave it off and do dithering in another program or sound-plugin (with more options for shape and quality). There are a range of dithers out there and imho I find Ozone's Intelligent II dither to be the best dither going. I strongly recommend you create your songs without any plugins on the master channel. Plugins even with 'light' settings can colour the sound distracting you from the job at hand: to make a clean mix. It's a bit like trying to paint a picture with sunglasses on.
Nice article, clean and to the point Also nice to add an example template. Will you cover rendering in your next articles? Note that when using internal samples and rendering with Sinc interpolation, then choosing the sample interpolation between "off" and any of the other interpolation methods, also has influence on this.
very nice read,thanks for sharing so when you have made a clean mix of your track,and you do the mastering yourself,do you bounce down the entire track as .wav from renoise or do you bounce the individual tracks as .wav??
great article, and very happy to read that more is coming. this is the stuff i do not know nearly enough about. thank you.
Vince V: I'll be covering Rendering and interpolation issues at a later date, but not likely the next article. I've a feeling the next article might be a little less geeky and a little more fun oriented. ErotiCore SteNch: You have two basic options: One is to do the mastering after the final mix by just using plugins on the master channel (still without touching any of the settings for the basic master channel settings that I explained). I'll cover this later in another article, but generally a few plugins ending with a master limiter (e.g. the Maximizer) will cover the task of basic mastering within Renoise. The second option is that you do a clean render of the final mix as a wav file. Then you can export that wav file to anything you like to do the mastering. I personally use Reaper because I like the ease of the routing options in that program for mastering, but there's no reason why you couldn't load the wav into a blank XRNS and do your mastering in Renoise. Depends on your aim too.
Nicely written, thank you! Can you recommend a saturator / soft clipper for use on individual stems? I recently begun using the native distortion in minimal amounts to smooth peaks over , it seems to work in the sense that it acts as a crude compressor, but you can only go so far before audible distortion occurs. I'd like something that can squash away and not produce the distortion, and without the 'fiddliness' of a compressor. Looking forward to reading more :)
Sure moshak, there are many options out there. And yet again, this will be another article I'll write in the future. For now I'll point you to some nice freebies that I like to use: Blockfish has an awesome saturator in it which you can tweak the back panel options of to get the 'tone' right. It's a downright awesome compressor for a freebie too. The other I'll point to is Bootsy's TesslerSE saturator. It's pretty good at giving a hyped up feeling to drums and basslines that need some 70s bite to them. Use both these in subtle ways and you'll get some more channel-level cohesion.
An awesome write up, thaks for taking the time. I was almost on the right track but this has cleared everything up and makes perfect sense. I'll be able to compose with a lot more confidence now! :)
Good times! Cheers! :D
Hey, this is off-topic, but what theme is that? Where can I get it?
Should be in the standard theme pack, called mmd_gloaming. If that isn't spot on let me know and I'll upload the latest version. Nice and dark :D
Ah, there's an older version in the pack under the 'more' directory. Here's the updated version which is darker: http://m.thequietrevolution.net/mp3/mmd_gloaming_mbp.xml
Hey, thanks very much! Beautiful theme.
LOL @ "way too much booty". That seems to be a recurring theme with both headroom *and* some of these girls around the way.
Thanks for writing this one! I already worked like this but there are some good tips anyway :)
Hi, great article. This is what I was searching for. I have some questions. I automated a lot the volume in my tracks, and control the "final" mix of each instruments in the post-Mixer, because a lot of my samples and vst are very different in volume. In your example the post-mixer levels are all the same (o db), all your samples are not amplified, and you control the mix in the pre-mixer, I suppose this is because off in this particular case you dont need automation, I'm right? Or you consider "bad" use the post-mixer?
Afaik you can't automate the post fader. IF you could that would be great! What I usually do is put a Gainer DSP on the end of the chain and automate that. Gain changes throughout the mix are inevitable, and there is so much amplitude stepping in VSTs anyway that it's not worth loosing sleep over. I'll write more on mixing later, but if you can aim to get the sound you want with as few VSTs as possible then you're doing great for the sound quality. You'd be surprised how good just simple volume placement can be if done carefully.
Thanks for the fast answer! I believe that is right that the post-fader cant be automated, it deal with the differences between instruments/samples. The pre-mixer is there to be automated, if you want. Or like you wrote, automate a Gainer, great tip! Now I need more info on basic Mastering : ) An interesting article about Dynamics by Robert Henke (Monolake): "Production Notes II The music on this album has not been compressed, limited or maximized at any production stage. Why not? Once upon a time, music had dynamics. There were loud parts, and there were more quiet parts. Then came radio. In radio there is a technical limit for the transmittable maximum volume. As a consequence the average level of music with a high dynamic range is lower than the average level of music with a low dynamic range. The loudest possible music in radio is music where every element is constantly hitting the limit, music with no dynamics at all. Radio, and more recently mp3 players and laptop speakers influenced the way popular music is composed, produced and mastered: Every single event has to be at maximum level all the time. This works best with music that is sonically simple, and music in which only a few elements are interacting. A symphony does not sound convincing thru a mobile phone speaker, and a maximized symphony does not sound convincing at all. Monolake is about complexity, about details, about the elastic tension between beats in the foreground and textural elements in the background. We want to preserve that balance as much as possible in the final product and this is why the music on this album is produced without applying any compression. This is not a 'dogma', it is just a personal decission for this release. About the mastering: Mastering was done entierly in the analog domain, using a selection of vintage and high end EQs running at 'hot levels'. This impies there is a certain degree of saturation going on in very loud parts due to electrical characteristis of the tubes, transistors and audio transformers involved, but that's it as far as nonlinear behaviour is concerned. More (general) thoughts on mastering can be found here. http://www.monolake.de/interviews/mastering.html "
I have often wondered why renders sound different externally from renoise, will give this a go. Also snappy sounds have been an issue for me so will mess about with some compressor settings
God this saved me, I've been really at a loss with getting things sound decent and I've had the same problem with tons of stuff as Matt said, renders just sound completely different - and I mean REALLY different, every kick and bass note is distorted to shit. Could anyone point me out to a good mixing guide that deals with frequencies and such in depth? It feels like I'm the master of mud or something...
I can't write the whole mixing bible in one post, but I can say the next article I'm about to write deals with monitoring and translation.
This actually helped a lot! Thanks a lot. I'll be looking forward to your next posts!
+1 Confirmed.
awesome
Great article. Thanks! I guess I’m a little late though. It appears the example .xrns file is not here anymore... is there somewhere else I can download it?
Just a little FYI, anyone looking for a monitor controller do not buy the SM pro audio M-PATCH (any version). They are complete pieces of garbage. Almost like disposable piece of equipment. Now, if 10 of them came in a Kleenex dispenser, then maybe. Go for the Radial MC-3 if you want budget monitor controller.