Notes on Dusk

May 25 2021 Album Diary

Dusk is an instrumental EP about a tragic event from my life: the sudden death of my brother. It contains 4 tracks — 2 piano, 2 guitar — and is the first part of a twin EP. (The second album is called Dawn and will be released soon.)

Visit the official page to listen to the album. If you like the songs, I’d highly appreciate it if you bought the album, or simply spread the word!

In this article, I will tell you the “behind the scenes”: the process of composing, recording, and mixing these songs.

The Idea

The general idea behind the album is that of a “downward spiral”. The key of each successive song is lowered by a half note, and slowly changes from Major (usually regarded as happy, gay, active) to Minor (usually regarded as sad, dark, melancholy).

I eventually settled on this pattern: A Major => G# Minor => G Minor => F# Minor

Why? Because that made the guitar songs easiest to play. (Unlike piano, you can’t just move the notes up/down strings to raise/lower a guitar composition, especially if it employs lots of open strings and quick trills.)

Besides that, I knew I wanted to tell a story. I tried not to overthink this: my emotions basically wrote most of the songs. (This may sound a bit wishywashy, but it’s how most art happens. It just … pops into your head when you’re in the right mood, or when improvising behind a piano.)

I picked four moments from the night in which my brother died: the sirens from the ambulance, the trauma helicopter making one last attempt, the police stopping by to deliver the sad message, and the rough night that followed.

This is also why I decided to make this a “twin EP”, with the other being called Dawn. That one will be the exact opposite: an upward spiral, raising the key, the sun is rising after that rough night, a new beginning. I don’t want all of it to be sad and hopeless :p

1. The Sirens

Some parts of this composition were written down years ago. The only problem was that I didn’t even have a plastic keyboard for recording at the time, and I had other projects I wanted to create first.

So when the time came to make this album, I had basically already written 75% of it.

And that’s when I noticed something: these compositions sound a lot like the ones in my previous instrumental album. And that’s boring. I want to challenge myself, innovate, create new music.

That’s why I decided to start this song with an actual siren sound, which nicely fades into the music.

It took about 30 minutes to find several suitable sound fragments (of a siren). I raised these to the correct pitch and mixed them together to get what you hear now.

Listening to those sirens immediately inspired me! The first part of the song is an improvisation I did on top of those siren sounds, just because I felt like it. I thought, at the time, that I would just remove this later, or improve it. But when I played it back … it sounded perfect and I never changed a thing about it.

Recording the song was easy: I had already occasionally practiced the piece over the years, ever since I wrote it. I have an old plastic keyboard that is plugged right into my computer, which definitely has its pros and cons.

But I didn’t have a proper way to record an acoustic piano, so here we are.

Because of the horrible digital piano sound, mixing the song was a bit more work. The whole song is quite “percussive”: short, quick notes follow each other all the time.

What’s reverb? It’s a simulation of the “echo” or “acoustics” you’d normally get in a live room. The higher the reverb, the more your instrument sounds like it was recorded in a big church or hall. In soft parts, the reverb is annoying, because it sounds unnatural (our ears just know that playing a piano that softly should not lead to that much echo). But for loud, full parts of a song it’s critical to get any semblence of the beautiful real piano sound.

2. The Helicopter

I tried to simulate the whirring/rotating of helicopter blades … on a guitar.

How did I do that? The whole song has a “circular” plucking pattern. I keep hitting the strings one by one, in a fixed tempo, from low to high … and when I’ve reached the highest string, I immediately rotate back to the lower one.

(There are some variations, of course, but those are only present when switching to a different section of the song that should sound different. Otherwise the whole track becomes boring, stagnant, and even irritating,)

Besides that, the song uses a lot of dissonance. This creates tension and strengthes the sad atmosphere. If you listen to the first few seconds of the song, you’ll (probably) immediately hear that one note doesn’t quite belong. It’s not in the same key as the others. But because it’s just one note, and the others fit together fine, it doesn’t feel like I’m just playing a random set of notes.

(Or my guitar is out of tune … although that’s never a certainty, as I did not have the common sense yet to TUNE THE FRIGGING GUITAR before recording. Now I know better. But at the time, I was just like “hey, these few notes sound good to my ear, let’s play”)

The whole song is quite high pitched ( = it only uses the four highest strings on the guitar), with the exception of the middle part (starting around 01:37). This is the core of the song: it simulates a helicopter best, and is calming and fast at the same time.

However, all this makes the mixing process for this track reaaaally complicated.

Ideally, you’d have a good studio room (with good acoustic treatment, enough space), at least two (stereo) mics of good quality, silence, a proper audio interface, etcetera. But I only had a ten year old budget USB microphone and a workspace that simultaneously functioned as my mother’s sewing workshop :p

In the end, I played the song 4 times in its entirety. (It’s important to play it all in one take! For reasons I’ll explain below.)

Then I removed the noise with a standard De-Noise plugin. As to be expected, these plugins aren’t perfect: some noise remains, some nice acoustic frequences of the guitar are removed as well. It’s choosing the lesser of two evils.

Lastly, I added some reverb again. A guitar has a nice soundhole and is usually enjoyed from a nice distance. If you record the strings from close by (with your microphone), it sounds extremely dry and unnatural. So, acoustic guitar? Always reverb.

Below are some lessons I’ve learned from recording this song:

Tip 1: Make sure you’re comfortable

My USB mic did not have a tripod or anything. It just stood upright on my desk on a tiny “foot”.

But … if I wanted to point it at the guitar, and not get any environment noises (or, you know, my own breathing) in the recording, I had to play in a very awkward posture that caused many mistakes (and pain).

After a few tries, I placed some blankets and plush animals on the floor, put my microphone between them, and just sat on the floor with my guitar (in cross-legged pose).

If you’re comfortable, it leads to fewer mistakes, less fidgeting, and you can record for longer without tiring (or getting frustrated).

Tip 2: a trick for stereo guitar

If you only have one microphone, you can use this trick to make it at least sound a little more stereo.

Record the guitar, then copy the recording to a second track. Edit both tracks in different ways: one gets more reverb, one has several frequencies removed or added, one has a slight delay.

Now pan one track to the LEFT ear, the other to the RIGHT ear. Tada! Now it sounds a bit wider and more stereo!

However, this is a crutch, not an actual solution. It will sound much wider if you actually record the guitar with two microphones (or double-track it, if possible). Additionally, using the trick above introduces “comb-filtering” (because you have two nearly identical tracks fighting each other), which usually doesn’t sound great with acoustic guitars.

Tip 3: Have decent equipment

My old, rusty, broken computer crashed with no reason, causing me to lose all the original files for this song.

Fortunately, I still had the output and some other parts of the recording (which I’d done at a different date, or stashed elsewhere). It was just enough to finish the track.

Most importantly, though, yes, better equipment will improve your songs, but up to a limit

My bad USB mic was unable to record anything at a decent volume, without loads of noise, and without coloring the sound in annoying ways. Some things can be solved with mixing and effects … but certainly not everything.

Not long after recording this album, I bought better equipment: an audio interface, a solid microphone, a mic stand, etc.

These did not break the bank, mind you. My new microphone is only about 70 euros more expensive than the old one … but it’s a world of difference. The new one can record everything cleanly, with more than enough volume, and I can position it any way I like (with the stand). Because I have an audio interface, I can actually record multiple mics simultaneously, and I can listen to a metronome/backing track in my headphones.

So yes, buying at least decent equipment will make a huge difference. But a microphone that’s more than 200 euros? No, you don’t need it. It will be a minor difference, which you probably won’t hear anyway.

Tip 4: Don’t record guitar Direct

I had another semi-acoustic guitar that could be plugged directly into the USB microphone. This way, there’s no noise on the recording, because the sound is sent directly from the guitar pickup to the computer. Nice, right?

The problem: it sounds like a guitar being recorded from the inside. It’s an ugly, mechanical, shrill sound. Yes, in live situations, or in combination with a band it can be absolutely fine. But not for a solo guitar composition.

I tried to make it sound better for days on end. (Plugins, impulse responses, layering the guitars, etc.) But it was just impossible.

Always record a guitar with a microhpone, from a good distance, and experiment with the position and rotation of that mic. That will be the biggest difference.

Tip 5: Overdubbing does not work

I make mistakes. Sometimes I play the wrong note, my timing is off, or I play a note suuuper soft.

At first, I tried to overdub it. I’d simply record that specific note again and layer it on top.

This was a mistake. It’s near impossible to get the timing and hit right. But most importantly: notes ring after you’ve played them, so you can never just replace that single note, you’d have to replace all the notes that were being played/ringing/muted at that time.

If you really want to overdub, do so for full sections. Most songs have natural breaks in them. (Parts with a short silence, a chord switch, a tempo change, etc.) Record parts from one break to the next, and you should be able to swap between takes without problems.

But, if at all possible, just play the whole song in its entirety several times.

3. The Message

This song starts with a doorbell. It was not hard to find the right notes that would simulate the sound of a door bell on a piano. The only problem was that these sounds are usually in a Major key (because they have to grab your attention and punch through other sounds in your house).

Eventually, I decided to copy that idea, but switch the notes to Minor before the doorbell intro was done.

The rest of the composition is not like typical (instrumental) piano music. Because I’m actually a singer-songwriter, I always write with lyrics in mind, sometimes in other genres like musical theatre. Subconsciously using that attitude for this song, made it a bit like an instrumental version of a musical song.

Large parts of the melody have, thus, received actual lyrics to go along. You don’t hear these in the recording (obviously), but they were in the back of my mind whilst playing. (What does the text say? Well, as expected, they are the “message” that the police is giving.)

The other parts, however, were designed to NOT have such a clear singable melody on purpose. It made the final composition just sound better, more fitting to an instrumental piano track.

My greatest lesson? The performance matters more than you think The first take for this song, my thoughts were somewhere else. I played all the right notes … but the whole song just fell apart and didn’t sound interesting. The feeling of “communicating a sad message” just was not there. (A bad mix can be saved by a good performance, but a bad performance can never be saved by a good mix.)

When it comes to recording and mixing, this song was nothing special. Again, I used my old plastic keyboard, did my best to make it sound like a real piano (with loads of effects, experiments, double takes, etc.), and stopped when I thought I could do no more with what I had.

This is Tiamo from the future: After recording many more songs, I realized several things.

4. The Night

This song was, by far, the most difficult.

Some parts are near impossible to play on the guitar.

(Makes you wonder why or how I even write those parts. After ten years of playing guitar, default chords just become boring to you, and you start challenging yourself with the weirdest chord shapes imaginable :p)

At the same time, this track has several functions to fulfill: it has to continue the downward spiral, neatly tie up the EP, but also setup the next EP (Dawn).

I decided to place a Capo on the 2nd fret, but leave the high E-string open. Not only does it sound beautiful, full, and sparkly, it also enables me to play very high and very low notes. (This track goes both higher and lower than all other songs.)

(On top of that, Dawn shall also have a song with a Capo on the 2nd fret, which is literally the twin of this song. Hopefully, it gives a nice contrast.)

Before I hit that “record” button, I already practiced the song a lot. Still, I broke it into parts: the start, the middle part (where I play more chords and hit the strings harder), and the final bit.

And yes, the reason I separated that middle piece, was because of the lessons I learned from song #2: The Helicopter. It was just too loud and different to record in the same way as the other parts.

At first, the track ended with the same melody with which it started. But I had already done that too many times! And I had a better idea. This album tells about someone dying due to a sudden heart attack … so why not end with a heartbeat (that also slowly stops)?

That’s why this song ends with my best representation (on the guitar) of a heartbeat. I think this is the strongest ending to an album I’ve ever done. (Which admittedly does not say much, seeing how few albums I’ve made.)

At the mixing stage, the same old problem reared its ugly head: the song is very diverse and I had to make sure all notes (both high and low, soft and loud) sounded about the same. To do so, I used several phases of compression, and some automation here and there.

Bij het mixen kwam hetzelfde probleem weer naar boven: het nummer kent veel diversiteit, en ik moest ervoor zorgen dat alle noten (hoog en laag, zacht en luid) ongeveer gelijk klonken. Om dit te doen heb ik verschillende fases van compressie gebruikt, en hier en daar automation.

Hieronder leg ik heel snel uit wat dat betekent. Als je dat niet wilt weten, kun je het natuurlijk gerust overslaan.

What’s compression?

Each recording has a certain “dynamic range”. This range is between the softest note and the loudest note.

A big dynamic range is nice, because it leads to variation, interest, tension. But if it’s too big … the softest note will simply fall away and not be heard, or the loudest ones will be annoyingly loud.

As such, you often try to reduce the dynamic range. To keep some variation, without losing notes or being too aggressive. In other words: you compress the dynamic range.

Tip 6: Use compression in phases ( = serial compression) I used to throw one huge compression plugin on the whole song. This effect had to work really hard to keep the volume at an acceptabel level. Regularly, it had to reduce the song by 10dB … which is a lot. Too much, you might say.

Because you’re pushing the volume by that much every second, the track sound very uneven and unbalanced. It’s like someone is trying to sing into the microphone, but you keep pushing that person to the left, to the right, forwards, backwards. (This is also called the “pumping” effect. If you hear this, you know the compressor should stop doing so much work.)

Instead, try to make each compressor reduce at most 3-4 dB. If you need more reduction, use multiple compressors. The first should be used to basically balance out the sound, the ones after that should be used to shape the tone, attack, aggressiveness, etcetera.

(And in the end, throw a hard “limiter” on the song. This does nothing more than draw a straight line at a certain volume, and NO part of the song may cross that threshold. This ensures the song does not clip and distort. But it should be the last thing you add, and it should not do much work, because it’s “stupid” in the sense that it only has that threshold … and does nothing else.)

What’s automation?

When you change a fader for your track (in your DAW), it changes the volume for the whole track. For the whole song.

That’s good! That’s expected behaviour! But … we don’t necessarily want a certain instrument to be the exact same volume throughoug the whole song.

That’s where automation comes in. In most software, you simply draw a line going up and down, and the volume of the track is determined by the height of that line (at that specific point in time). The guitar is too loud in this short section? Draw the line downwards. It’s too soft, but only for these few notes? Automate it to be louder.

(For this specific song that line likes a cobbleroad where three land mines have exploded :p Usually the line is much simpler, or not needed at all! Vocals are heavily automated, but other instruments not so much. Which leads me to … )

Tip 7: Automation is very useful, but usually the problem lies elsewhere If you recorded your guitar way too loud, that will have distorted the sound, and you can never get it back by lowering the volume. On the other hand, record the guitar too soft, and cranking up the volume in the chorus will introduce loads of noise. Automation is a way to balance a track and steer the listener’s attention, not something to fix problems.

Conclusion

Despite the many mistakes during development and inconsistent audio quality (due to poor equipment), I’m very proud of this album!

Anyone who listened to it, told me it was (surprisingly) good. They mostly noted that it sounded unique and dynamic, yet was very calming and soothing to leave on in the background.

The piano songs are crisp and clear, but the guitar recordings are muddy and noisy. As I already said: better equipment has been bought and is being used now.

I also learnt a lot about “mastering” with this album: making sure all tracks on the album are consistent in volume and tone. (And making sure this volume is the expected volume on music services like Spotify.)

On top of that, I learnt a lot about the best spaces to record, good placement for microphones, and how smart effects will make the track sound bigger and larger. But there’s a limit, and that limit is based on how good your performance and original recording was.

(For example, the 4th song has a part that goes expectionally loud. I’m just furiously hitting chords there. My mic didn’t like that and mostly recorded the lowest notes and the “ticking” from my nails onto the strings. Afterwards I searched for the precise frequence of this “tick” and lowered it using a plugin, while reasing the volume of the higher melody notes around it. It’s certainly not perfect, but it DID make a huge difference in clarity.)

Until the next album!