How I Approached Reharmonization for the First Time

[#music] [#chiptune] [#lsdj]

December 02, 2023 - 9 minute read

I have wanted to try reharmonization for a while now and finally got the inspiration to sit down and do it. Reharmonization is an exercise where one takes a piece of music that already exists and rewrites the chord progression while keeping the melodic identity of the song unchanged.

This was my first go at it and I didn't really know what I was doing, so I winged it. But I had a lot of fun and am pretty happy with the results! Here's how I approached it.

The Song

I reharmonized the chorus from my song Welcome To The Woods. The chorus starts at about 1:08.

The main parts directly transcribed are like this. The song is in E mixolydian, so every D will be natural instead of sharp.

The song as originally written.

In the bigger picture of the song, the chorus represents a moment you might have when you're on a nature walk in the woods and see something move, get a little nervous, only to find out it was just a squirrel. A moment of very slight tension in an otherwise very chill song.

With that in mind, I think of the chorus as the "suspense part" and then the "relief part".


The first two measures of the chorus. We start with some 7add9 chords. They might seem fancy at first, but I definitely wasn't thinking about the music in that way when I originally wrote it. I think I was just noodling around on a piano. All I did was stack triads.

I played an E major triad and then connected a B minor triad on top, and those two combined can be written as E7add9. The weird voicings in the transcription arise from the semi-randomly programmed LSDJ tables I used.

Same deal with D7add9. It was born out of an A minor triad and then a D major triad.

I think the only reason I ended up with an A minor, with the non-diatonic C natural, was because I used the same table as the previous B minor chord to voice it. If you look closely you can see the intervals in the A minor chord are identical to those in the B minor chord.

Happy accidents like that are one of my favorite parts about LSDJ. There are a lot more parts of my music that are only there because I was too lazy to program another table. :-)

In between the chords, there is a little call and response idea going on. The chords are the call, and the response melody just bounces back and forth between D natural and G sharp. It's a tritone. Spooky! But it's the tritone that lives in the E7 chord that's already being voiced, so in context, not that spooky.


At the end of the 2nd measure, with the help of an A major triad, we come crashing back down to an E major chord and everything is resolved. The D natural in the spooky tritone melody becomes an E, making the melody echo the E major chord. (The transcription in the above video is wrong there... oops.)


This suspense-relief cycle repeats almost verbatim, with the spooky tritone melody raised an octave, and the final relief part allowing a longer melody to finish the chorus.

I really like how in the first relief part, the melody gets interrupted by the E chord in the fourth measure, and then in the second relief part, there's no interruption. The melody can finally say what it was trying to say and connect the disconnected parts you've already heard. It's really satisfying!

How I Approached Reharmonization

Now that I've explained the song I was working with, I can go into how I approached the task of reharmonizing something for the first time.

My approach to reharmonization was to deconstruct the song into what I considered to be its "core identity" - the smallest set of notes that I would still consider to be the same song. My thinking was that stripping it down to almost nothing would give me a blank canvas to rebuild it from. The approach worked out really well and I would do it again this way.

Here's the identity:

The "core identity" of the song

Not really any surprises here. Framing it as call and response again, we keep highest notes of the call chords and also keep the responses. Not shown in the transcription is the rest of the relief part. It follows the same principles.

The other thing we keep is the A chord before the relief E chord. This is the song's main motif. It's repeated in the verse sections too.


In the verse sections, the motif is an F# to an E instead of an A to an E. Looking at it using scale degrees (here as roman numerals), in the chorus, the motif is a IV to I, and in the verse it's a II to I. If you know a little bit of functional harmony you might recognize that IV and II are both subdominant chords and these are just two different voicings of the same functional idea. A simpler way to explain it might be to simply notice that F# minor and A major have multiple notes in common (A, C#).


Now that we have a skeleton of the song, we can put some meat on it again.

The First Idea

This was my first go at it:

My first reharmonization idea

The first thing that came to mind when looking at the identity was that E is only a whole step away from D. What if I put a little D# in there to make a nice chromatic walk? And it worked pretty well!

Adding that D# chord meant that now the chord progression advanced every 2 beats instead of 4, so the second measure needed a chord change. I took the D major triad that was already present in the D7add9 and added a B below it. That makes a Bm7 chord.

I changed the A chord that leads in to the relief part to an F#m/A chord. This was actually just born out of a mistake in transcribing the original. For a while, I thought the chord in the original song was F#m/A instead of A - and I liked it! So once I found out, I just kept the F#m/A in my back pocket for the reharmonization.

I added a Bm7 to the end of the relief part. This makes for a little more drama and motion leading into the repetition, makes for a nice context change under the last response melody, and is a fifth above E so it makes the repetition loop nicely.


The D#maj7 contains a G natural in the bass, and the response melody contains a G#. This is a minor 9th and according to conventional music theory wisdom, this is 🤢 THE BAD INTERVAL 🤮.

Well... sounds fine to me! The response section is supposed to be dissonant anyway. Plus, since the two ends of the interval are contained in different instruments, it's not even that noticeable. In fact, I didn't notice it until I transcribed it.

You could also view the descending E7 -> D#maj7 -> D7 as secondary dominants where each dominant is using the tritone substitution of the chord a perfect fifth above it. D#maj7 was originally simply D#7, but I cut out the 7th (C#) because it clashed with the D in the melody. I'm not sure if you can really call them secondary dominants if every step is a half step though...


I liked this version, but I was also curious what else I could do with the tritone in the response melody, so I made a second version.

The Second Idea

My second reharmonization idea


Instead of the chromatic descent of the first idea, I took the tritone in the response melody and figured out what chords it would fit in. I ended up with probably the most obvious answer to that question, A#7. Another way to look at it is that A#7 is the tritone substitute of E7 and the notes were already in the original E7add7.

The next thing I did was try to connect the D7add9 chord to the F#m chord. I noticed since dominant 7th chords like to move down in fifths, G7 might be a natural chord to follow D7. Bonus points because the D in G7 matches the melody. Then I noticed that G7 could be considered a secondary dominant to F#m as the tritone substitution of C#7. Funny how that kind of just worked out. I also tried to voice the G7 in a way that the voice leading to the F#m was nice.

That G7 is interesting, but I'm not sure if I like it. Too many perfect fifth cadences always starts to sound cheesy to me. Another thing I don't particularly like is how it changed the phrasing. To me, the there should be pretty clear boundaries between the call part and the response part of the call and response ideas in this song. Putting the G7 secondary dominant before F#m kind of glues the end of the response to the beginning of the next call.

Oh well! It's still interesting, and this whole thing is just an exercise anyway. A negative result is still a result!



If you were curious:

  • I figured out the initial ideas with a piano, transcribed them in musescore 4, then programmed them in LSDJ.
  • I made the playback videos by simply screen recording musescore 4 while Sameboy played back the music out of frame.
  • You can download the .sav file containing the original song on my Bandcamp. There are some free album download codes on this website somewhere. :-)