Notes on 'Keeping the Sunset on my Side'
Recently I released my first ever singer-songwriter EP called “Keeping the Sunset on my Side”!
It was quite a long journey. Many parts were recorded … then re-recorded once I realized some grave mistakes I made … then re-recorded again when I realized plugins and editing could not save everything.
In this album diary, I’d like to explain the process behind the creation of this album, the biggest mistakes I made, and what I learned from those.
Why this album?
I originally planned to create another album first. It was literally a collection of the first 16 songs I ever wrote in my life. Most of them were quite simple, so I thought it would work out.
I managed to record large chunks of the album, before realizing the task was too great. There were too many songs, too varied, too challenging, for me (with almost zero experience in this department) to record and publish succesfully.
So I said to myself: do something simpler first. An EP, not an album. Only singer-songwriter stuff, with just a guitar and a voice, just like you’ve always done.
Fortunately, during the recording of that other album, I’d written many songs. (Whenever I had to test a microphone placement, or my acoustic treatment, or just wait for my slow computer to boot, I would improvise on the guitar. Which often lead to these songs.)
I said: “7 is enough! Let’s draw the line here and record this as an EP first!”
And that’s how “Keeping the Sunset on my Side” was born.
(Fun fact: the actual title for this album, and the song that contains it, was only created after I’d already recorded most of the other songs. Until that moment, it was just called “EP #2”)
The original idea
I don’t like explaining (or analyzing) the music and lyrics to death, but that doesn’t mean I simply threw together some songs and don’t want to talk about it.
The title of the album is about my tendency to be a bit melancholic, to only see the upcoming dangers and how our society is absolutely failing to combat them, to be prepared for heartbreak before I’ve even fallen in love. I keep the sunset on my side. That’s what all the songs are about, one way or another. And the fact that it ends with “Guess what? I’m still keeping the sunset on my side”, should tell you about the current state of affairs.
Most of the important melodies and phrases had already been written before I even decided to record this EP. The rest was completed and filled in by letting this theme guide me.
(For example, there’s literally a demo on my computer where I sing “Somewhere in the dark, the mighty king still runs, somewhere in the shadows he is sticking to his guns” It was recorded a few minutes after I came up with that melody and that line. The rest of the song was built around that. And unsurprisingly, the title of the song is “The Mighty King Still Runs”)
Step 1: Recording the guitar
Stereo mic-ing (plucked guitar)
If a (plucked) guitar is the only accompaniment, you simply have to record it in stereo. Otherwise it sounds extremely weak and soft.
I experimented with many different microphone placements.
- “Room mic”: one microphone up-close, the other at the back of the room (high in the air) to catch the sound of the room
- “Overhead mic”: one microphone is up-close in front of the guitar, the other is hanging over my (right) shoulder, capturing roughly what my ears would hear when playing the guitar
- “X-Y”: two microphones on top of each other, one angled to the left, the other to the right
- “Spaced pair (pointed)": two microphones to the left and right of the guitar, quite some distance away, but both pointed at the 12th fret
- “Spaced pair (straight)": same as above, but both mics just point straight ahead
These are my conclusions:
- “Room mic”: makes the initial recording sound great … but is absolutely useless afterwards. It doesn’t create a proper stereo image, and its effect can be easily recreated using a reverb or delay plugin.
- “Overhead mic”: sounds nice and crisp, the two mics complement each other well. However, it’s only a very small stereo image, as both mics still essentially record the same “plane” of the guitar. (This might be solvable by only recording the high strings with the overhead mic, and the low ones with the other. Something to try out in the future.)
- “X-Y”: a clear stereo image, but not very wide. Easy to setup, no issues (as far as I can hear) when combined.
- “Spaced pair (pointed)": a much wider stereo image, but you need to be careful about the placement of the microphones. If they are out of phase, or one is significantly softer than the other, you can get more cons than pros with this setup.
- “Spaced pair (straight)": as expected, this gives the widest stereo image, but also leaves a kind of “hole” in the middle. You’re literally recording the extremes of the guitar, but not the center, which causes you to lose some body, but offers a good space for the vocal to sit.
In the end, I think the “X-Y” is the safest bet. At least when you’re just starting out with recording, and the songs are simple. It provides a nice stereo image, but doesn’t bring any dangers with it.
Double tracking (the rest)
However, that doesn’t mean I could just record one guitar performance (with two mics) and call it a day.
If you’ve been hearing the same plucking notes for a minute, you probably want something extra for the next verse/chorus. Similarly, on some of the bigger songs, I did want a wider/fuller stereo image than X-Y could give me.
For all those situations, double tracking is your friend.
(This simply means: record the same part twice, then play them at the same time, but with one panned to the left and the other to the right.)
Every song on the album has at least one part where I added some chords this way. Usually, these come in on the second verse or chorus.
Then, when the bridge or final chorus comes in, another double tracked part comes in.
This way, I keep building and building the sound, with more layers and a wider image, until the climax of the song.
Here are some ideas for extra guitar parts (on top of, for example, a main plucking pattern):
- One-strum chords => at every chord change, just strum the new chord once. Double track that and it sounds great.
- Arpeggios => play the songs, but pluck one string at a time. If you use a predictable pattern (for example up<->down), you can double track it to great effect.
- Alternative chords => all chords have different “positions” where they can be played. Even though it’s the same chord, it often sounds different. Play these at the same time (as the regular chords), and they sound like MEGA-CHORDS.
- Muted strumming => if you need some more flow/rhyhtm/continuous force in the song. Just mute the strings and strum the guitar onto the beat. Watch out that this strumming pattern doesn’t interfere with the other chords, because then it becomes messy.
In a sense, the original “X-Y” recording provides the main driving force of the song, but all these double tracks make it sound like more than just a bedroom recording.
Less is more
There you have it, the great lesson number one: less is more. How predictable!
Many songs on the album only have (at most) 3 or 4 parts on top of each other. And even then, it’s rare that they all play at the same time.
This wasn’t always the case. I originally recorded many extra chords, melodies, arpeggios, alternative chord voicings, etcetera.
But it was just too much
I tried to make every guitar part sound cool on its own, which made the final track overwhelming. And if you try to double track that, you’ll find it to be an impossible task.
No, the parts should work together, to support the vocal. I ended up cutting many parts, only leaving a few notes here and there that worked well, or only leaving the parts that lined up nicely.
Transitions and Guidance
Instead, I learned to focus the guitar parts on these two principles: transitions and guidance
Let’s say you have a soft, mellow verse that climaxes into a wild chorus. At first, I would play both parts, and cut-n-paste them together.
And guess what? It sounded … fine, but there wasn’t a clear climax, there wasn’t any tension being built and released in the chorus.
No, I needed an actual transition.
If you listen to the album, you’ll often hear that an extra guitar comes in before the chorus. It’s just a set of chords, or a note being played louder and louder, or some muted strumming. But it comes in on the last few beats, builds and builds … until it stops when the chorus explodes into action!
Of course, I still have much to learn, but believe me: the tracks sound much more dull without these simple transitions.
And then I realized: “hey, these parts are actually more about guiding the listener to where I want them to go, rather than building this huge musical tapestry all the time!”
I learnt that it’s best to have one focal point at each moment in the song, and one only.
Whenever a vocal is happening, that’s the focal point. Otherwise, you have “empty space” that you could fill with some “ear candy”. You could play a nice little jingle high on the guitar. Or you could switch up the chords to something the listener hasn’t heard before. Anything that could guide the listener’s attention while the vocal is silent for a moment.
But it also means that, while the vocal IS making sound, you should not try to place loads of backing tracks underneath it.
If you listen to the song “#4 Don’t Pretend”, you’ll hear this quite clearly. When I’m singing, there is only basic accompaniment (guitar strumming + double-tracked chords). But during the long pauses between phrases, a different high guitar pattern plays every time. In fact, it alternates between the left and right ear, because that felt fun to do.
Now imagine what it would sound like if these weren’t there. Or if these high jingles were EVERYWHERE in the song, even during all the singing. Chaos, I tell you.
First and second draft
It took me a week (one song a day) to record all the guitar for this album.
Or that’s what I thought.
When I’d recorded all the bits and pieces, I thought: “Nice! Cross that off the list, now it’s just a week for vocals!”
No, that’s not how it works. Upon listening several times, on different devices, I could hear the mistakes and the possible improvements.
A track might be too chaotic. Two parts might not be nicely double-tracked. There might be too much empty silence in some parts of a song.
In the end, I had to do a “second draft”-week, where I recorded all sorts of extra guitar parts, and edited or removed many of the originals.
This is normal. Nothing is perfect the first time. But don’t let that discourage you: just create your best first try, give it a rest (for a week, or a month), then come back with fresh ears and make the necessary improvements.
Besides all these major steps, there were of course many small things I’ve learned after recording so many stuff.
- Compression and layering bring out all sorts of noise. Remove any “silence” from takes, and use fades at the end of clips (or on transitions).
- I’m not the great musician I thought I was :p When you record yourself, you suddenly hear every tiny mistake, every (accidentally) muted string, everytime you suddenly played too softly (or too loudly) without a good reason. I’ve become a much better guitar player, just by recording myself a lot.
- High notes are very audible. Adding too many high patterns/chords from a guitar will overwhelm a listener. Even a single tiny mistake in those frequencies is enough to ruin the whole listening experience.
- At the same time, low notes tend to fall away … but not because they are mixed so well, simply because they turn into a sort of “mud” at the bottom of your mix. Try to keep the bottom end clean: only one part should play there, and it should not play notes very close to each other. (General rule: the higher you go, the more freedom you have when it comes to pitch and timing. In the lower registers, a strong bass line that follows the main beat is usually all you need.)
- Tune the guitar. Tune it after every take. Often it won’t be needed, but don’t let that fool you! I’ve had many a recording sound “off”, and when I checked the guitar tuning, one of the strings (usually A, D, or G) was suddenly wiiiildly out of tune. Just after 10-20 minutes of normal playing!
- I feel like condensor mics are more suited to Spanish guitars (classical guitar, nylon stringed) than Western guitars. When I use them on my western, it very quickly just sounds … harsh and brittle. As someone once said to me: opposites attract. Condensor mics are quite bright, so they like the mellow sound of a Spanish guitar. If I have the money, I would rather buy a different mic for my Western (maybe a ribbon mic?)
Step 2: Record the vocals
Oh boy. You thought the guitar part had many steps? Well, get ready for the ride of your life!
We’re going to talk about GETTING GOOD VOCALS.
You see, I had already committed a deadly sin by now. Vocals were “part two”. They shouldn’t be — they should definitely be “part one”.
From now on, I will record the vocals first, then do the instruments. This decision was made because:
- People listen for the vocal. It’s what they want to hear, it’s what they connect to, it’s what makes or breaks a song to them.
- The vocal is the loudest, most audible, most important part of each track. If it isn’t great, if it’s even slightly off-pitch (too many times), it ruins the whole song.
- At the same time, vocals are the most dynamic and imprecise instrument. A good, emotional vocal will NOT have perfectly quantized timing, nor a consistent volume, nor perfect pitch. It’s almost impossible to be free and dynamic with your singing, if you have a strict backing track (and metronome) that you’re trying to fit into.
- If I play some wrong chords somewhere, I can easily re-do that section, or maybe do some quick editing to fix it. Heck, I’ve re-double-tracked many guitar parts on the album just because I thought “why not? It only takes 5 minutes” If I sing all the wrong notes however … yeah, there’s nothing saving you now. (Unless you really like the sound of autotune.)
I ended up recording all the vocals for this album (at least) three times. Why? Let me explain …
The Great Problem: Pitch
At first, I recorded the vocals the traditional way. Headphones on, backing track in your ears, sing into the mic.
When I listened to the solo vocal, it was GREAT. When I listened to the vocal on top of the guitar support, it sounded … a bit off, but I thought that was normal when you hadn’t mixed the track yet. (A closed-miked voice without reverb always sounds a bit “dead”, for example.)
So I recorded everything this way, sometimes doing multiple takes if I screwed up, but usually happy with the vocals themselves.
But then … I tried and tried, but the vocals just would not sound even moderately acceptable. What could be the issue? If I listened to the vocal in solo, I thought it sounded awesome, and didn’t hear any weird noises, or echos, or frequencies. But combine that with the mix, and it was horrible.
I did many experiments, which ended with me writing two articles about “Singing with Headphones”.
The conclusion? Whenever I sing with headphones, I am consistently flat. The reason the vocals sounded great in solo, was because all the relative pitches were correct, and my voice was clean and relaxed. However, I had to tune the vocal upward by 50 cents to make it the actual pitch of the song.
Long story short: I had to learn a setup to sing on-pitch into a microphone, and re-do everything.
Double tracking voice
This is the main reason why many vocal parts on the album ended up being “doubled” (sometimes clearly audible, sometimes just a soft effect to help the vocal stand out)
The main recording you hear is the last take I did, which is actually at the right pitch. All the supporting recordings are my old takes, shifted a 1/4 note upwards.
(Shifting those old recordings made them 95% perfectly on pitch. However, it introduces some artefacts and removes some of the natural sound, because, well, you’re literally changing the notes. So it wasn’t usable as a main vocal, but quite good as a support.)
At other times, I tried to re-do parts of a vocal, but found that I couldn’t combine the NEW with the OLD. I couldn’t remember what my old microphone setup (or singing technique) was, and it just didn’t match my newer recordings in terms of tone. Adding the new recording as a double was the only way to mitigate this issue.
(In fact, I recorded many vocals with two microphones slightly apart, because I thought it sounded better. Guess what? The only reason that sounded better, was because the recordings cancelled each other out, making the pitch issues less noticeable. So it was a bad idea, and I removed all those extra tracks and just recorded my voice with one microphone afterwards.)
Lastly, I figured out that it’s very important to make the double tracks line up almost perfectly. (Especially if they are triple tracks, or more!) In real life, that “huge” sound you get from a choir or just a group of people singing, often comes from them singing the same notes at slightly different pitches or timings. So I thought that it would be fine to be “not perfect” with the doubles.
It wasn’t. It sounds like a confusing mess if multiple voices are saying the same things at different times, for three minutes straight. So I ended up spending quite some time making the extra voices fit perfectly, by cutting, moving and stretching where needed. However, in the future I’ll make sure I just make the doubles match when recording, because it saves a lot of time, and reduces the chance of a double take sounding bad when you’ve edited it.
(My main issue was: random note extension. Whenever I had to hold a note, I apparantly did it for a random amount of time, because those transients rarely matched between two takes. Now that I know this, I can keep it in mind. In fact, I’ll probably just play the melody with my piano first, and then use that recording as reference for the vocal. It should ensure more consistency.)
Because I already had so much trouble with getting a single proper vocal recording, I didn’t dare do anything fancy with the double tracks. All of them are just the same melody sung again. (Which is a missed opportunity, but still the right choice, I think.)
Only by the end of the album did I gather the confidence to try a simple harmony. This ended up being the outro to the album. You can hear three voices:
- One singing the main melody (which is mid-range for my voice)
- One singing an alternative melody that is a bit lower
- One singing an alternative melody that is a bit higher
Together, I think it creates a nice harmony, much nicer than just playing three near-identical takes of the main melody. (Which I tried. Believe me, I really tried to make this sound good and full. At one point, I had 5 takes playing at the same time. But it’s obvious in hindsight: how can the sound be full, if all you hear is the exact same melody multiple times? That just makes it louder.)
Of course, I didn’t just do this off the cuff. All three parts were created as a MIDI, just by clicking my mouse on the MIDI editor, then moving notes around whenever I thought something sounded bad (or could be better). Once I was satisfied with how they sounded together (using a standard Grand Piano VST), I recorded them.
I literally recorded them sitting behind my desk, using my terrible laptop microphone. Why did I do that? Well, remember how I said I had huge issues singing with headphones? Those issues, partly, stem from the fact that I’m 100% used to singing live in a room, playing guitar at the same time. In my recording studio, with acoustic treatment, singing into a mic, I’m just not able to hit the right notes at the moment.
Sitting behind my laptop, I could get the full harmony in 1-2 takes. I used a noise filter to remove any ugliness, edited the vocals to match perfectly (which is essential, as I explained above), and that was it.
Why do I tell you this? Because it brings me to the next lesson: getting a good performance out of a vocalist (me, in this case) is more important than having an amazing room or good gear. Without any editing, and without any preparation, I could sing a harmony that sounded better than many other vocals on the album … which were recorded with my good microphone and edited heavily.
In the future, I’ll probably use this same technique when I hear my vocal takes are lacking. Because it seems silly to waste hours trying to get a good recording behind a good microphone, if you can get the perfect recording at the first try when sitting in a different room/setup.
Remark This does not hold for many, or any, other instruments. The voice is a single-point instrument and very focused (sound goes straight forward). So even in my bad room, with my bad laptop mic, I won’t pick up that much reverberation or outside noise (as long as I sing close to the laptop). If I had recorded guitar this way, it would sound terrible, as it’s a much larger instrument that needs some space and will interact heavily with the room.
Keep it simple
And yet … after all this talk of doubled voices … I realized I was “overproducing” this.
Many of these songs are just simple singer-songwriter melodies. Just because I can now enrich them with all sorts of fancy stuff, doesn’t mean I should. I think the next album will be simpler, with fewer voices, and fewer layers of guitar. Keep it simple, keep it small, if that is what the music asks for.
And if I really want a bigger vocal (e.g. during a fuller song, or a chorus, or a big climax somewhere), I’d probably not “double” it, but actually write a harmony for it. It just sounds way better and is way more fun to sing and edit, because each part is different. (You’re not just singing the exact same melody over and over and over.)
Remark I also realized that my way of playing guitar usually produces a wide frequency range. (I like playing chords with all strings, and I like switching up my chord shapes to non-standard stuff higher up the fret.) This means that I was often battling myself when I tried to add more layers of guitar, because all of it was filling the same frequencies, especially with the voice. So, in the future, I’ll either try to get a full arrangement with just one guitar (using my usual way of playing), or purposely split the frequencies into something like: low (bass notes), mid (standard chords), and high (high variations on chords)
I hate the sound of autotune. I hate artists that use it on everything, as much as I hate musicians that can’t actually make music.
And yet …
In my articles about “Singing with Headphones”, I explain all the issues I was having, as well as my health issues that make singing (consistently) even more of a hassle. I had to make a choice. Spend another few months, hoping I can improve my vocal technique enough to provide perfect recordings? Or do I apply subtle autotune on the worst offenders to finish the album now?
I chose the latter, because I found the results better than expected.
On many recordings, I only had to tune a few important notes that were off by (again) 1/4 note. I used Reaper for that, with the ReaTune plugin, because it’s very fast, intuitive, and doesn’t change the natural sound of the voice (if applied subtly). A few weeks later, I can’t even tell which notes I autotuned.
Interestingly, when I tried to (or asked others to do so), they pointed out parts of the vocal that were not autotuned. Apparently, my voice has the tendency to sound like autotune by itself on certain phrases :p
(This is especially true on phrases that very quickly wobble around a few pitches. For example, take the start of the verse in “#3 There Is A Need”. Many of those starting notes sound a bit autotuned, but they’re not.)
So I settled on a middle ground. Most of the vocals are completely natural, which (I think) is also how they sound, as you can e.g. clearly hear my vibrato (instead of it being flatlined by autotune). But whenever I listened and thought “aaai”, and couldn’t fix it with a few extra takes, I tuned it.
(And as stated before, all second and third voices were NOT tuned like this, only completely shifted upwards when needed.)
Understanding the Lyrics
Whenever I played someone a first version of these songs, their responses were ALWAYS about the vocals … and always conflicting:
- I tried to hear what you said, but couldn’t understand many of the lyrics! I think the vocal should be louder.
- I think the vocal was too loud, I’d like to hear more of the guitar and make them blend together! I think the vocal should be softer.
As it turns out, both were right, and neither were right :p
By default, I set the vocal volume quite loud, so I can hear everything I’m doing. (And because, by now, I’m very afraid of my recordings being 1/4 note flat again >:( )
But that means it uses the softest notes as the bare minimum. I want to hear those as well, so the volume is raised until I hear even a whisper.
You guessed it: this makes the loudest notes way too loud. Causing the vocals to be both unintelligeble at times, and annoyingly loud at other times.
- Volume automation
First, I went through the vocal and made everything more equal in volume. Not perfectly equal: that would be boring (and too much work). I just want the gap between a soft and loud note to be much smaller than it is on a raw recording.
Then I added a compressor. I aim for at most 3-4 dBs of reduction. If I need more, I use a second compressor.
By this point, I like to do the “vocal first test”: lower the volume until you can barely hear your track. If the vocal is still audible, whatever part you’re listening to, it’s loud and consistent enough.
Now add some reverb (maybe delay) to give it more space, make it sound “larger than life”, and blend it with the rest of the track. Do note, however, that reverb “softens” a sound (as you push it further back from the listener), so check the volume again after enabling this plugin.
Once I had compressed the vocals, I could usually lower its volume considerably, and it would still be audible, intelligeble, without drowning out the guitar.
Side Note I’m not that experienced with frequencies and the EQ plugin. I know the basics, and I will use it to get rid of low-frequency rumble or a muddy guitar sound, but that’s where it stops. When comparing my vocals to professional ones, however, I noticed that they had MANY more high frequencies. So I started adding a high-shelf boost (around 6-7 kHz) to almost all vocals, and I think it helped.
And that’s how we got here.
I had to figure out a way to sing on-pitch, a way to naturally fix the worst notes (and blend voices together), and realize the essence of a few default plugins, to get a professional sounding vocal.
Pfew. What a journey. I learned a lot, I was frustrated a lot (because of all the mistakes I was making and the amount of work I had to re-do), but it brought me here.
Do I think this album is amazing and worthy of toppling the charts? Of course not. I have a long way to go. It’s decent at best. I have to force myself to publish it now, otherwise I would keep re-recording everything and trying to make it more professional.
I published it, and will now continue with the next EP, and then the next one, and so on until I hope to be very good at this “music production” thing at some point in the future :p
Notes per Track
As always, I have some (hopefully interesting) things to say per individual track.
1. Keeping the Sunset
Writing: The two parts of this song (the intro with “keeping the suuuunset on my side” and the rest), were not written like that. They were two different melodies that I simply stitched together. Both just came to me when I was improvising with my guitar. The chords aren’t anything fancy. The lyrics also came easily, as it is a topic dear to me: my frustration with being forced to wait, and wait, and wait until my life could actually begin.
Guitar: The main guitar you hear (playing the rhythm + chords) is a single recording using two microphones. In hindsight, it might have been better to just double-track that. It would have given a wider stereo image, and because it’s just a bunch of chords being played on the same fixed rhythm, it wouldn’t be hard to make the two takes line up.
Because it’s just a recording of a single Spanish guitar, I felt it lacked the strength needed for this song. So I recorded two more guitars:
- One that’s just continually stomping the rhythm on the lower strings
- One that plays the chords, but higher, and with some embellishments in between (when there was some dead space)
I still think these guitars could sound much bigger and stronger, but hey, hopefully I’ll learn how to do that with more experience.
Vocals: The following will be true for all tracks: the vocals were recorded multiple times. The first one was off-pitch. The second one was mostly on-pitch, but inconsistent, making it impossible to double track (or fit the timing of the guitars). The third one was mostly perfect, with only a few notes tuned.
Mixing: Once I had the guitars, and the vocals, I realized the verse and chorus flowed into each other a bit … boringly. Too smoothly. Like there wasn’t a transition at all.
So I added two guitar parts (one on the low strings, one on the highs) that just walk up/down a scale to transition into the chorus. Nothing amazing, but it’s already much better this way.
The transition from the intro to the rest used to take longer. I just played a few “in-between chords” (to get from C Major to C Minor). But I plucked them too softly, with too much silence between the notes, so I ended up cutting a lot of that silence and heavily increasing the volume.
Similarly, there used to be an extra measure before the solo starts. It’s gone now, because, well, it was just awful and didn’t fit the groove of the song.
2. Make me Feel Wanted
Writing: The second song I wrote for this album. I remember waking up with this melody, telling myself to write it down … and then forgetting it, until it popped up again when I went to bed. This repeated for several days, which was both annoying and confirmation that the melody was quite good (because it stuck in my head and I really wanted to use it).
Finally, I went to my studio, and laid down a quick demo of the chords and some nonsense lyrics. And somehow, whilst improvising the lyrics, the line “that I cannot be seen with you in my arms, safe and warm, everywhere we go it’s maaaagic, it’s magic dear” came out. And I remember listening to that demo and just feeling the truth and impact of that line. (In fact, I tried to use that original take for the vocals, but it was just too muddy and mumbled.) So that determined the feel of the song.
Guitar: One guitar, recorded with two microphones, played in three takes. Nothing else, because the song doesn’t need it.
Vocals: This was the first time I tried to sing into the mic while softly playing some notes on the guitar, to help me get the right pitch. This also means that the vocal you hear was just a single raw take, with only a few parts edited (either slightly tuned, or made louder, or something similar)
In fact, this was such a success, that I decided to use that technique for everything else on the album. Without guiding notes? I’m consistently 1/4 note flat. With guiding notes? 90% of the takes were perfect. And to my ears, the “bleed” from the guitar into the microphone isn’t audible at all. But maybe that’s because I’m still quite untrained as an “audio engineer”.
Mixing: The default stuff: level out volumes, add reverb to make it all sound better and larger, De-Esser on vocals, some saturation/crunch to make it a little warmer. (I find these recordings with really clean mics, in a clean space, to be quite sterile. So I’m still experimenting with ways to add back that warmth. That cozy feeling that a cute singer-songwriter melody should have.)
3. There is a Need
Writing: Ah, the first song I wrote. And I didn’t even actively write it. I had just completely redesigned my bedroom (for the third time) to get better acoustic treatment, and a space for recording that was easier to access at all times.
So I sat down. I had to test it. What better way to test it, than to get a crappy guitar and play chords as loudly as I can? (I wanted to hear if there was any reverb/echo, if the sound was being colored, if I liked playing in there.)
It turns out I did like it a lot, because this song just came rolling out of my fingers (and my voice). A few tweaks to the lyrics, an outro, some structural changes, and it was done.
Guitar: As I said earlier, I’m still not sure how to make a set of acoustic guitars sound any bigger. I don’t know if it’s even possible, and am considering adding more/different instruments for other albums. In the end, you hear me playing the basic chords, and then two other tracks (one with rhythm and one with the same chords/melodies, but higher) to fill it up during the big sections.
I still have doubts about these backing guitars. Because the melody and the lyrics move so quickly, there isn’t much “dead space” to fill, so you’ll often hear the background guitar doing fancy stuff while the voice is also singing. In the end, I think the volume levels I set (and stereo spread I used) make it just possible to hear everything without being overwhelmed. But I’m not sure.
Vocals: Somehow, the chorus just suits my voice. I sang all choruses in a single take, without a single wrong note, and when I analyzed it, most of the notes were even dead-on. This was a happy moment for me :) Unfortunately, no such luck with the verses, which took some more tries.
Like with the guitars, I hoped to make this sound bigger and more … aggressive, I guess? As the song progresses, more and more double tracked vocals are being added to the chorus. I mean, it does sound nice, and it does sound fuller than the single voice from the verses - but it’s not enough, in my opinion.
A song like this, I guess, requires some more instrumentation and a larger band to properly get across the message.
Mixing: Nothing special, although leveling out the vocals took a lot of work, because the song is so dynamic. (Almost every phrase had a few words I sung extremely loudly, immediately followed by a few I tended to mumble.) But in the end, it’s worth it, because it makes the song much smoother, more professional, and easier to understand and follow.
4. Don’t Pretend
Writing: I was playing around with Maj7 chords on guitar, simple as that. I was moving them around, up and down, until I got something nice.
When I recorded this as a demo, I immediately jumped to the phrase “don’t pretend”. And when the verse ended, I decided “feel like I’m in the flow, let’s immediately continue with a chorus!” And so the chorus was born.
This is one of those shorter, more cute and “obvious” songs. The message is clear: don’t pretend you’re someone you’re not, don’t lie to yourself, don’t hold back on stuff you’d enjoy.
Guitar: I experimented with a different technique here, using one mic on the guitar, and another back in the room to get the echo. It didn’t work out. I ended up with basically one guitar sound. Which was nice, but not stereo, so very small and weak when played on headphones (or good speakers).
In hindsight, again, I could’ve double-tracked these chords.
But I didn’t. Instead, I kept the original recording in the center, and played some supporting figures. Those were double tracked and panned left and right!
Why? Because I found I really enjoyed these songs when they started with a single small guitar in the center … and then suddenly opened up when more guitar parts on the sides appear in the mix. Just a nice transition that works well with these songs.
Vocals: These vocals are actually quite messy, though I don’t think you could tell just from listening to the track. It’s a combination of three things:
- The original vocals, but shifted 1/4 note upward (so they are on pitch)
- Re-recorded vocals, which I sang over my speakers. (So I just played the guitar accompaniment through the speakers, in the null zone of the mic, and sang along with that.)
- A select few re-recorded vocals using the latest technique (play the song at the same time on my guitar)
The chorus is doubled because it just fit well. The chorus has these long drawn-out notes, with big open chords being played underneath. In those cases, a single voice feels too small.
The “don’t pretend” is doubled for the sole reason that it was the only way to mask how these different recordings were stitched together :p (And, also, because no matter how often I sang “don’t pretend”, it always sounded wimpy or bad or stupid for some reason. Something about those words … or those specific pitches … )
Mixing: When I sent the first version to other people, I had not re-done the vocals (so they were quite off-pitch), and only had my single guitar recording (without effects or extra layers).
The feedback? “Sounds nice and soothing … but that guitar is somehow very harsh and aggressive?”
Yeah, that’s the problem with close-micing guitar (with a single mic). If you don’t add reverb, if you don’t push it back a little, it just sounds like … strings slapping on wood :p The lesson here? Don’t judge your guitar sound before you’ve added some subtle, natural reverb.
But interestingly, the vocals didn’t seem to matter to anyone. The lesson here? As long as you’re close enough, and have a nice tone, vocals can be wildly off-pitch and nobody cares.
5. The Mighty King Still Runs
Writing: Another one of those moments where I just wanted to test a new setup (in my studio), or had to wait until my software had booted, and I suddenly played this song.
It’s quite rare these days that my guitar has a capo on it (when improvising). Which is the reason this is the ONLY song on the album which uses one. But I think I need to use it more, as it does generate different sounds, and leads me to different ideas.
The theme was clear from the first moment I spoke those lyrics. Even though someone might battle with illness, even though someone might be beat-down or afraid to step up, it doesn’t mean they’ve given up the fight. At any moment, they will rise above the ground, and the world will follow.
Guitar: Here’s a funny story.
I played the background guitar for this song. All in one take, I was perfectly in time (I had practiced a lot the last week), it flowed nicely, I was proud of myself!
And then I realized … I’d put BOTH my recording channels on the SAME microphone.
Which means I only had a SINGLE beautifully plucked guitar recording, which was useless if I wanted that wide stereo sound.
The lesson here? Check that you send the RIGHT microphone to the RIGHT channel, before hitting record!
I was so frustrated by it, that I put off re-recording the thing until the very end. Needless to say, I never got it that perfect again, but I managed to do the job.
The support for this song is a very quickly strummed pattern full of many different chords. This was rhythmic and full enough, that I didn’t see the need to add anything else. (If anything, it was already too busy for me to handle.)
I only added extra guitars (playing the same chords, but in different variations or octaves) during the transition bits (“reeevooolutions”). They were really necessary there.
Vocals: As usual, the first recording session was horrible, then I re-recorded it and got something okay, then I retuned (or generally edited) a few bits to make it all the same quality.
Especially those transition bits were dreadful, so I spent some extra time figuring out what each voice should song, and making sure those were perfectly on-pitch. To be honest, I still think they could have been better. But at the time, it was my first try doubling voices that sang something different, and I was proud to get anything usable at all!
Mixing: Again, I mostly saw a need to add “transitions” between important parts. Chorus to bridge. Out of the bridge, on a high note, to the last full chorus.
In that same vein, I noticed that some work with “automation” can go a long way. Increase the volume with 0.5dB in the chorus. Increase the amount of reverb subtly during the solo. Minor changes, but they really help navigate the listener’s ear through the song, instead of making everything just the exact same volume and energy.
All the rest, both guitar and vocals, have some compression, some reverb, some doubling, but that’s all.
6. Walk Awhile With You
Writing: This song has a lot of key changes and weird chord progressions. Some chords I didn’t even know the name of, before I looked it up.
Unsurprisingly, it was hard to write a good melody over those chords. The first part of each verse (until that last “someday soon the world is gonna change, and you won’t change with it”) was already locked in. That was the melody that inspired the song! But everything else was figured out by trial and error, again by sitting with a guitar on my lap for a few hours, and trying everything I could think of.
I think it ended up a beautiful song, with beautiful lyrics, that’s really different than anything else on this album (or I have ever written, actually).
Guitar: Well … that also means it was a pain to record and mix :p
The guitar plays, plucks and strums a lot of difficult chords and changes. I ended up recording one main guitar (with two mics, correctly this time!). But I found it too static (it was just the same soft plucking all the time), so switched to chords for the last 1/3 of the song.
But that turned out to be a too-sudden shift, so I eventually double-tracked some plucking back into that last 1/3!
And to give it all more texture, and to add some nice high frequencies, I improvised all sorts of patterns on the high-strings throughout the whole song. Like with “There Is A Need”, it was a constant battle between “variation” and “clarity”. I ended up removing many of those extra guitar parts, especially when there were already multiple voices singing, to keep it more streamlined.
Vocals: Okay, if you already have trouble singing the right pitch (with headphones on) … try doing that when the song switches keys all over the place.
I had to write down the exact notes I was singing, and play them on my piano before every recording, to keep myself in check. And even then, this song required more tuning than I’m comfortable admitting. Some sections were 100% fine. And then another section appeared, and my ears decided to fool me again, and I thought the key was a half step lower!
I re-recorded what I could, stitched something together, and made sure to get some nice doubled vocals to choose from during the chorus. But still …
Mixing: … this one required the most mixing and editing. Just to get everything in tune, to line up vocals with the guitar, to line up the guitar with other guitars. Meh. It was a bit of a mess. And I hate that, because I think the song is beautiful and I want to do it justice.
(When I gave the first version to my dad, he (hesitantly) said: “The first songs [from the album] are nice and easy to follow. Those later songs though … they are quite hard to follow. Could it be … could it be the vocal is off-pitch?” Yes, dad. I hadn’t mixed and edited those yet, and this song in particular sounded HORRENDOUS at first.)
Nevertheless, I managed to pull something from that dumpster fire. And probably because the song is so diverse (in terms of chords, melody, accompaniment) I actually found myself listening to this one more than any other from the album!
7. On my Side
Writing: It all started with the four chords from the verse. I still don’t know what you’d call them, but they sounded really nice and different and … spicy :p
Once I had those, it wasn’t hard to find similarly spicy chords for the chorus. In case you’re ever stuck: for the chorus, just pick similar shapes, but much higher on the guitar! Ta da!
After playing around with these weird chord structures for a few days, I settled on the final one (you hear on the album), and decided to find a melody to sing over it. The album is called “Keeping the Sunset on My Side”, and this was to be the final track, so I gravitated towards the lyrics “it’s shaping up to be a cloudy day”.
That settled the theme and the rest flowed automatically.
(The benefit of choosing weird chords, is that it’s harder to sing anything over it. I usually have to try individual chord notes, find common ones, stitch them together. But because that is such a focused process, you usually get a single nice end result out of it.)
The lyric “Finders keepers” just popped into my head, because it fit perfectly with the chords. In a sense, once I had that, I knew the rest of the song would be great and easy to finish. (I did finish it the same night.) Sometimes you just feel that.
Guitar: You wouldn’t expect it: a single guitar playing all these weird chords. Double-tracked guitars that start with a simple strum (whenever the chord changes), but move to a quick arpeggio/plucking at the end of the song. Then another set of double-tracked guitars that are a rhythmic driving force.
This song is very similar to “Don’t Pretend” in that way. And that’s no accident: I liked the techniques used, so I re-used them on a song with a similar tempo and strumming pattern. But that’s also the reason why that song is #4, and this one is #7 (the last). If I’d placed them right after each other, as I originally had, it would probably sound repetitive (or like one big song that never ends).
(The plucking at the end of the song, before the outro, was recorded as an alternative to the chords. I ended up using the chords anyway, but I liked the plucking so much, that I kept it in. Even though I don’t fully hit a note two times. Trying to keep it real, you now? :p)
Vocals: Again, some parts just worked for my voice. You could wake me up at 3AM to sing “Finders keepers” (on that melody), and it would sound cool.
The chorus was also dead-on with each take, with the exception of the “daaaaaaay” long note. I would sometimes go to the note below it, by default.
The biggest issue with that note, was the fact that it’s one word, but it goes through different pitches. (If you listen to the song, “day” has a sort of wobble in the middle.)
Well … if you try to double (or triple) track that, it is VERY hard to line up that wobble so that it sounds good! After many tries, I had to basically tell myself “okay, the wobble occurs early, after exactly X beats, so stick to that!”
The other big issues was, again, the key change. “Finders keepers,
My voice kept pretending there was no change, and tried to sing notes from the original key over the new chords. Again, I had to write out the exact notes being sung, and practice them specifically (with a guitar in hand), to get rid of this “habit” and get some good takes.
(Funnily enough, I didn’t even recognize this at first. I knew the original vocal take was a bit off-pitch here and there, but I didn’t know I actually completely sang the wrong note. And I even wrote the whole song! This recording process has just taught me so much about music, about singing, about pitch, it’s unbelievable. Now when I sing, I can hear much more easily if I’m doing the right thing or not, just because I spent weeks analyzing my melodies and sung pitches.)
Mixing: Above I already explained the creation of my “3 voice harmony” for the outro. Yeah, that required most of my mixing time.
The rest was, at this point, standard fare. Although the vocals needed heavy automation again to get the volumes somewhat balanced. I learnt I have the tendency to randomly sing notes VERY LOUDLY, before returning to a soft mumbly vocal. I’m trying to get rid of this, control my dynamics more (and more consciously). But for now, I have to go in and spend at least 30 minutes editing the vocal so that you can clearly hear everything, but no note suddenly blasts your head off.
And then we add reverb, compression, some saturation, a De-Esser, all the good stuff.
I debated (with myself) about the outro a lot. Especially the transition from the actual song to its outro. I tried writing an actual transition, but that felt overblown. I tried removing many of the layers in the outro, but then it felt extremely small - certainly not a worthy ending to an album!
So I settled on a middle ground. Some of the extra guitar layers I recorded, were either removed or heavily simplified. The vocals are just the 3-voice-harmony, nothing else (so no “doubles” on that). And the transition is created by playing a few “in-between chords” to get from the Dsus2 to the C.
When I say “in-between chords”, I usually mean one of two things:
- I move the bass note step by step. In this case: D -> D/C# -> C
- I use the Dominant chords to step between keys. For example, you could do D -> D7 -> G -> G7 -> C
If any of this seems like magic, look up some music theory (about dominant chords, tension, and inverted chords). Or just experiment on the guitar, that’s what I did for most of my life.
Pfew, the longest article on this website yet. Hopefully it gave some interesting insights in what I learned, how I wrote and recorded this album, and the lessons I’ve learned.
I probably left out hundreds of small lessons. You just learn so much when you try to professionally record something for the first time ever.
I know, for a fact, that my current workflow is 100x better than what I tried 3 months ago. (Better mic placement, checking if everything’s correct before recording, better arrangement, better setup for getting good vocals, etc.) But most of these things are a habit by now, subconscious. I couldn’t tell you exactly what I changed and why.
So, if anything, if you have music you want to record, just go and do it. That’s how you learn the most. Not by reading articles like this, or watching YouTube videos on mixing. It’s a truth I also find uncomfortable — because it’s much easier to just sit behind a screen and let someone else tell you all the secrets, than actually doing the work yourself and failing 9 out of 10 times — but nevertheless true.
And those were my notes on Keeping the Sunset on my Side. Let’s go to the next EP!
(As with all my music at the moment, I’m keeping the option open to do a non-acoustic version, or a remastered version of all songs. I’m sure they could all be recorded and mixed way better, so don’t see this as my final work that will never be touched again. But it’s a starting point.)
Just before launching this album, I was able to get some feedback from people who were (somewhat) professionally working as a producer. These were the main points:
- Play the guitar a bit less aggressively, push it back a bit more in the mix.
- Bring the vocals to the front => really just make them much louder
- Sometimes, the voice isn’t 100% in tune. It’s okay to use subtle autotuning, or aggressive tuning on background vocals. In fact, all professionals do it.
- Try to tell more of a story with the songs, by building towards something. Now most songs are (mostly) the same instrumentation, the same volume, the same intensity throughout.
- (The song “Make Me Feel Wanted” would probably be nicer if it was player more slowly. The current version feels hasty.)
All points I could agree with! As such, I managed to convince myself to do one more mixing day and see where I can get. These things were changed last-minute:
- Some guitar parts were either lowered in volume, or compressed more (so that the initial impact wasn’t as harsh). For future albums, I’ll see if I can move the mics back more, and pluck the strings more consistently.
- The vocals on all tracks were made at least 1-2 dB louder. In the end, they might be too loud again on the final album, but hey, it’s hard for me to find a balance on my first album.
- Many doubled vocals were autotuned. And they were right: you can’t hear the robotic artefacts it creates (because it’s just a background vocal), but the song overall sounds much nicer. (Of course, these vocals were already “fine” to begin with, it won’t work on something that’s completely off.) If there were still problems, especially in the main vocal, I did extra takes on the spot trying to nail those few notes and nothing else. Usually, I got something better (and more natural sounding) within 5-10 takes.
- The last two points … yeah, a little late for that. In fact, I already tried to do this, but just didn’t have the knowledge at the time. The next few albums, I will certainly play around with this, making conscious decisions about the arrangement and “story” of each song.
In the end, these tips don’t suddenly make the album 100% perfect and professional (obviously). But I do think they help a lot, for such small edits, and they gave me some more perspective. Now I have a better “ear” for the correct volume of a vocal, which is certainly a great thing to have.
(Looking back, it’s baffling to think that I thought the original very quiet vocals were okay! They were not! Simply comparing one of my tracks with a professional one, immediately showed that their vocals are much louder and more prominent.)