If you ever tried to create music, you will know this feeling: you open a new empty project, and you stare at it… what should I do first. I know, I have been there. Or even worse, you are feeling in the zone, super creative, and you are filling up your project with great sounds and effects, only to be completely helpless and demotivated at the end because you created an uncontrollable wall of sound. I have been there too 🙂
I hope this very long post will help you structure your way of working a bit more, so we can avoid all of that. This is how I (usually) work.
Practice, practice, practice…
First of all: no one can create good music from day one. Some will reach their potential a lot sooner, others (like me) may take years to get there :). Doesn’t make it less fun though! So one thing to keep in mind is: you need to practice over and over again. This is how I practice:
- I also read a lot, and watch a lot of tutorials, about music production. There is no need to discover everything by yourself. Don’t reinvent the wheel. Learn from the best (easiest thing in the world today because of YouTube). Don’t be afraid that this will cause you to become less creative: it won’t. You will still do your own thing, but you will become more efficient at it.
- When I create a new song, I don’t this track by track. I work in phases, which I will explain below, and I do this as quickly as possible. Quick decisions are good: you keep your attention, and your intuition will guide you what works for you. There is a balance you need to find though: do things too early, you will loose a lot of time for nothing. Do things too late, you may have to rework a lot of things.
- Kill your darlings. If you started off with an idea, and it doesn’t seem to work out, that’s ok. save it and delete it! On one of my last songs, I worked on a cool EDM melody… but it no longer fitted the idea of the song. That’s ok: I deleted the melody, kept it for later, and finished my grooving song I was working on. That melody is still there and I will create yet another track from it, no harm done.
- Then, most importantly: finish all of your tracks! I will always, while writing music, have 2 or 3 moments where I am unhappy of what I did and where I just wanted to trash the project. DO NOT DO THIS. On every occasion, I learned a lot by finishing the track. At the end, you will know what you should have done different in the beginning. And most of the time, it still ended up ok.
Build your template, and improve upon it
When you create a new song, you need to start with your template. You need to create one, and improve it over time. With every song, you learn what works and what not – you want to include that knowledge in your template for the next one, and use it as a “cheat sheet”.
Here is an example of what I built so far, and what works for me – you can follow it, but I would recommend you build your own.
There are three main sections in my DAW:
- The instrument layers, what you see on the left. In this example, I am describing a bass sound that uses 3 synthesizers to build its overall sound: the low-end signal, which is usually side-chain compressed with a click or kick track. This wont distort the sound as much, but creates a lot or room for the transient of the kick
The sound volume at this level is up to 0dBFS (full scale, peak)
- The instrument bus, where I do my processing of all the layers of that instrument together.
I may occasionally add a creative effect in here, other than a delay or a reverb. Delay or reverbs only exist on my fx channels, unless I need to animate my reverb for one specific instrument in a track.
The sound volume at this level is between -18dBFS (pads) up to -6dBFS (kick). This is also where I start doing track volume automation and panning, for instrument fade in/outs. Automation is used on the filter, as well as the width plugin, to change the sound color (high pass / low pass) and stereo effect, to increase and decrease tension.
- The group bus, where I do my processing of all the bass sounds together. The channel strip includes an eq (with limited options), gate, compressor low/high pass filters and additional gain staging.
The sound volume at this level is relative to the other instruments. This is where I do my overall track volume automation, to change the focus and balance of the song. I target an average of -12dBFS here.
- The main bus where I do my master bus processing and top-down eq/balancing
- The listen bus where I do all my metering and where I gain-stage the track to sound louder and less dynamic before it goes to my speakers.
- the sanity block is an assistance tool I sometimes use, especially when I have been working on a song for a few hours: iZotope neutron, for example, can auto-analyze a sound signal and propose improvements. If the improvements that neutron suggest are radically different, it usually means I am doing something wrong. I may need to stop, rest and revisit the project later. I usually don’t end up using the suggestions this tool does though – but it is a good indicator of things are setup ok or not.
Here is an example of my actual template. It lists a lot of plugins that are disabled, and I will not use most of them at the end – but they are a reminder of the steps I need to think about when I am creating my sounds: from midi, to instrument layers, to instrument bus… all the way up to my speakers.
When I work within these blocks, only process my sound where I need to and already start thinking about compression, EQ and saturation along the way, and make sure it sounds loud enough on my speakers as it would sound at the end while I am making these adjustments, I will need to do less corrections when I am eventually mixing my song.
Step by step
My template contains a general structure of how my mix should be setup. But it does not include any instrument present. In fact, all these busses sound the same. There is only one instrument I use at the start: my piano (I like the NI Noire plugin :D).
Sketch your idea
I usually start writing chords, and a basic melody, starting with 4 bars and extending it up to 16 usually. The goal of this sketch, is to write down the main idea and to compose everything you will need in that group of piano bars. It will force you to:
- write something that sounds good from a melody point of view, without any influence of your sound design
- focus on rhythm to make things more exciting
- write something that sounds in harmony together
- use the full scale for your sound (because you cant put another note where there is already one ;), which will avoid sounds clashing later
Make sure you spend enough time in this phase. Don’t settle for your first idea, and work it out in a bit more detail. Change the rhythm or add syncopation, add lead notes, some ear candy, change the volume/velocity of your notes, etc..
I will write a dedicated article on this topic alone in the upcoming weeks.
Decompose your idea into parts
Once you have that blueprint, you want to start extracting parts of it into multiple tracks. This is where I start moving parts of the melody to what will become my lead sound, the pad, the bass, etc.
For the drum bus, I define two basic rhythms, using only a kick and a snare. You may notice that, in these two first phases, I try to keep things as simple as possible. The goal is to produce something quick and fast, and validate your idea as early as possible so you don’t loose more time later if your basics don’t work.
I will share with you how this may look and sound like when I am working on a song.
Yup! It’s that basic 🙂 But that is where we all start, and build from, piece by piece…
Find your instruments
Now finally, we are going to look what instruments will play out basic musical parts.
You should not work on your full sound design yet – only choose the basis of the sound you will go for here. We will make it sound nice later. If you do this too early, not only will you loose a lot of time working on something you may never hear in the mix of your sound – you may end up hurting the mix altogether because your sound may become too complicated.
Once you selected your instruments, before continuing, make sure you gain stage all your sounds: with all your fades leveled, adjust the gain of each track so that your sounds are in balance.
Once you have your basic idea worked out, it is time to do your initial composition. Spend enough time here as well, as this (and your sketch) are usually the most important parts.
Your song should evolve over time, increase in tension, decrease in tension and give the listener the feeling that the music is going somewhere. Make sure that not every verse, chorus, drop or hook sounds the same. Try to do a small change every 4 to 8 bars, and make sure you change at least two instrumentations every transition.
I will write a dedicated article on this topic alone in the upcoming weeks.
This is also the moment when you start adding your vocal tracks. Depending on the genre, story and structure of your song, you can do this before or after you arrange your instruments.
Humanize your track
By now, you will have your basic composition, and it should start to sound a bit more like a song. It wont sound great yet though, and even a bit robotic. This is a good time to start humanizing your tracks.
Merge together bars, and start creating small modifications here and there. Notes that are sometimes just a bit different from the regular pattern that happened before. Add a few extra octaves here and there, or add an extra instrument for and extra sonic dimension.
And, more importantly: change the velocity and timing of your midi notes. Add some swing, apply “humanization” to your melody to make it sound more like someone is actually playing those instruments.
Compose extra melodies
This is a good time to start adding extra melodic effects, such as arpeggios, risers, small melodic fantasies, counter melodies, etc. Use it sparingly, and where needed. Don’t exaggerate.
Remember: your melody needs to remain simple, repetitive, recognizable… but it also needs to change, grow, be exciting… there is a balance between boring and too complex 😉
Build your grove
At this moment in time, we still have a simple kick/snare drum. Now we need to build out all our drum patterns and create our main groove. There are two ways you can do this: you can either play your drum with your midi pad or keyboard, or you can use a pattern-based groove box. For realistic drums, use your midi. For efficiency and more electronic grooves, you may be better off creating patterns. It is easier to compose drum tracks if you use patterns, but it will sound less real.
Don’t forget to variate your velocity/volume, and put notes off-beat to add some more swing to your midi notes.
When I finish my overall drum tracks, I will already start adding some compression to control the dynamics. Right after that, I will again gain-stage all my sounds throughout the mix so that they sound in balance, with the drum dynamics pushing trough in your sound.
Ideally, by now, you will have a ballpark of around 8dB of headroom and an overall dynamic range of 9dB up to 6dB for louder electronic club and dance music.
Produce and automate your sounds
By now, your song could be more or less done. Or in other words: it should be good enough for most people to enjoy. If it is not, it may be best to stop here and re-iterate over what you have done so you can make it better. The remaining steps are important to make it sound great, but nothing you will do there will be able to make the song better on its own.
The first step in finishing your song, is making sure your instruments sound really good. This means, adding layers to key instruments (not all) to accentuate different parts or to “decompose” a sound into multiple parts we can control individually (like I do with the bassline).
Now is also a good time to start controlling the frequencies of your instruments and EQ them to make each part stand out, use compressors to control the dynamics of the sound and to start automating certain parts of your effects (filter high/low passes and the width of each sound) to create extra tension and movement in your sound. For example: make sure your bass goes away before the drop. Or try to make your lead and vocal sound more mono during the verse, and more stereo during the chorus.
Yes, again 🙂
Add extra ear candy
Your song is almost finished. But it still needs some ear candy.
I will be adding extra sounds, vocal chops, lyrics that just say “Hey”, extra backing or detuned vocals… all at certain places in the song, to keep the listener’s attention or to let them know, something is going to happen.
Create your transitions
Another form of ear candy are sounds to announce transitions. For example: risers that increase the tension just before the drop, a swoop or drop to relax the tension, reversed cymbals and other percussion elements to keep the excitement… Next to that, you can also produce and alter certain sounds to increase the tension even more: a riser made out of a reversed reverb part of a vocal, or a gated/tremolo effect on an instrumental part.
Don’t overcomplicate things though. The goal is to make the transition predictable to let the listener know what is happening. Although it is also a lot of fun if you can trick the listener into thinking a drop is coming, to then only postpone it for a bit. 😉
Yes, again 🙂 But just a little bit, to make sure your new sounds fit 😉
Up until this time, my song is still mixed in mono. I did not apply any panning so far. It is done on purpose, because gain staging in mono is easier and it would help avoid phase issues from the start.
All the sounds are now in our song, and they should sound acceptable by now. It is time to give them more room:
- keep bass and core rhythmic sounds in the middle
- pan certain items to the left or right (not all the way). be very picking, don’t start panning everything: it needs to make sense. If you pan too much sounds, the stereo effect will disappear, because the listener wont be able to hear the difference anymore
- key elements now need to become louder (less balanced) than the rest
- items that need to disappear in the background need to sound softer, with some room reverb to make them go away
- add some additional reveb and echo where you want the sound to loose focus, or where you want to create more rhythm. Only do this if it makes sense and if it matters (and only there)
With everything panned in stereo, you will need to rebalance your sound again. If not, parts that are panned will sound softer on non-stereo systems, such as smartphones, clubs, Bluetooth speakers etc.
Finish your mix
We are almost done. The very last thing I do, is add additional volume automaton on my busses. I am really talking about a few dB here and there. For example: my drum buss will increase by 1.5dB during the drop. My basslines will be about 3dB softer during the verse, etc…
Master (as good as you can)
The last step that remains, is mastering our song. Before I do that, I let it rest for at least a week. You need to reset your ear and forget what you have heard before you can mix and master your song again. Ideally, you want someone else to master your song, so he/she has a fresh pair of ears. But even if you do that, you still want to give a good product to the master engineer, so that your intent is there.
The last things we need to do:
- apply some saturation and tape effects, to give the song a bit of a crisp or push
- shape the overall sound to what we like
- then, verify the tonal balance with some reference tracks
I usually use a 5-multiband compressor for this. I set a very very soft knee, a low ration, and I will use the signal “gain” knob to tweak it just +/- 2dB (max).
- finally, run it through a series of compressors. If I did my mix right, I only need to compress it a bit (a few dB here and there).
Final mixing and mastering tips
A few last tips:
- Don’t wait until the end of your production before you start limiting or compressing your signal. It is important that you hear how your instruments may sound like when they are mixed and compressed at the end. If not, you will have to rework a lot of your earlier EQ and balancing work.
Also: don’t compress everything – only where you really need to :).
- Once you finish your composition, humanization and production of your audio – it may be a good idea to bounce or render your instrumental tracks. Not only will you save processing power, you will also save your project in case an audio plugin stops working later on. Before you do, make sure you save your project under a different name (eg: newSong-version6 or something, there is no limit to how many versions you can store).
- You don’t always need to use side chain compression. You can also use a LFO tool or volume shaper. It will help you save processing power, but it also allows your DAW to use more CPU cores. If you start side chaining your busses early on, your DAW cannot do as much parallel processing anymore, which may create a performance bottleneck at the end.
- Before I start working on a song, and before I decide to publish it, I always play it trough once. If I don’t hear anything that is off, or that I want to change, then the song is done 🙂
This was my way of working, or at least at the time I was writing this. I tend to update it after every song.
Do you have a different approach, let us know in the comment section! 🙂