What's new

Using Synth V for Songwriting #2: Painting Better Melodies

Reid Rosefelt

Well-known member
I’ve always wanted to be able to write better melodies, and Synthesizer V has helped me more than any other piece of software.

While I could have seen the shape of my melodies if I wrote them as notes on a staff, or as MIDI notes in a DAW, I never did. I just sang them. Recently, I compared some of my old melodies with the ones I loved—the catchy, hummable and memorable ones, The difference was obvious. The great melodies moved all over the scale, up and down like a snake. There were leaps, from one note to a much higher one. Some-WHERE over the rainbow. Take ON ME, etc. There’s a book called “Switched on Pop” that has a chapter that reduces all of Taylor Swift’s melodic writing to a formula the authors call the “t-drop,” descending a short distance, and then descending a BIG distance.” I think this is ridiculous, but it is undeniable that Swift’s songs are brimming with vocal leaps and, IMHO, this is what makes them so catchy.

There was very little dynamic movement like that in my melodies. No interval leaps at all. Why was that?

I decided it was mainly because I wrote my songs while singing them, playing my guitar. Right off the bat, I was limited to my paltry vocal range, which is in the typical octave and a half-- if I’m in the right key. Notice how nobody sues Elton John and says, “I wrote ‘Tiny Dancer”--not you.” Right. Go ahead and sing it if you wrote it, Mr. Bigstuff. I’ve studied my favorite Brian Wilson songs. As everybody knows, he goes way up there, and this, IMHO, allows him to create the kind of leaps that tug at your heartstrings.

I’ve also come to believe that singing to a guitar progression sometimes held me back. If you have a fun guitar progression with a lot of movement, you can sing a few notes over it, and it still sounds like something is happening. The not-very-exciting melody interacts with the chords to create contrast and musical energy. The combination works, I think, in a way not unlike the way a pedal does on a Bach fugue.

Synth V has solved all of these problems for me. There is no longer any issue of vocal range. I’m currently working on an album where I’m hiring a lot of singers. I can write as many octaves as I want—I only need to hire a singer who can handle them. Not only can I write these notes, I can also hear what it sounds like when a “singer” sings them. This is very freeing.

Synth V enables me to write like a singer. I can incorporate all the tricks singers use like into my compositions before they go out to the singers. I have a song where one word steps down five notes. With Synth V, you are a contestant on American Idol. You can do runs and falls and melisma. Anything you can think of. The great part is that when I send these guide tracks to real singers, they take my melodies even further.

After I put my melodies into Synth V, I have a clear visual display of their shape. Does it have a lot of up-and-down movement? Are there a few leaps? Rhythmic interest? None of the above? What I put into Synth V is just a first draft. The actual work begins when I start experimenting, painting my melody, shoving it around like clay. Does this line sound better if I take it up rather than down? How about up and down? Or down and up? The scale is my friend. My only enemy is stasis, going nowhere. And I never quit. The melodies are the same as my lyrics. I keep trying to improve them.

After I am happy with my first verse, problems can come up with a later verse. While a line in verse two might have the right amount of syllables, it sounds weird when I put the lyrics into Synth V. I have found that when everything is perfect with a lyric and melody, it goes into Synth V very quickly, with no hassle. But sometimes there is something that is discordant. Sometimes it's just bad pronunciation, but just as often it's because the accent is on the wrong syllable. It's pronounced fine, but it still sounds off. And I work and work on these little things, until I stop and try changing the notes. In every case that solves the problem. For some unknown reason, the second syllable in a particular word needs to be lower. (I do think this is the kind of shift good singers do instinctively, but synth vocalists don't know how to)

One thing that happens often in modern pop music is that they have a lot more melodies, a lot more sections. But it’s not necessary to write more sections to create more melodic interest. I have a song called “The Naked Truth,” that only has two sections and an outro, as far as the accompaniment goes. But the A-part and B-part are all sung different ways as the song progresses. I don’t have the words to call what each verse-with-a-new-melody is, I just work to create a flow from the first note to the last in the outro. “The Naked Truth” is a very old song. I started writing it in 1984. I’ve been working on improving the lyrics over the years, but it was only finished when I put it into Synth V and transformed the melody. I should mention that I took a song I’d been singing for decades and turned it into a song that I couldn't sing to save my life.

I am certain that all my best melodies have been written since I got Synth V. And I have a process that serves me on every new song.
 
Last edited:
As you've said, the temptation while singing to a guitar or piano is to stick with a comfortable range. That's not necessarily bad, but it is limiting.

When creating melodies at the piano, the tendency is to go with muscle memory, and use patterns that lie under the fingers. It also results in a more pianistic melody.

When creating melodies in SynthV, I find I have the opposite problem as you. My melodies have too many chordal outlines and skips. So I'll go back through a melody after I've created it, and ask "Can this be simplified?", and the answer is almost always "Yes!"

Being able to see the melodic curve also means I can see if the melody has too many or too few repeating elements. This is especially helpful in making sure my chorus doesn't sound like the verse.

On my last song, the chorus sounded like a verse, not a chorus. I took a break and listened to some songs to remind myself what a chorus should sound like. Then I dragged the existing chorus past the end of the song (so I'd have a copy if I wanted), and drew in a new one. I adjusted notes until it worked. Easy peasy, problem solved.

I usually write the lyrics after I've created a backing track and melody. Writing lyrics is the hardest part of songwriting for me anyway, but doing it this way helps ensure that the end result will be musical.

I'll give myself a fair amount of flexibility on adjusting the duration and pitch of notes when adding lyrics. My focus is on intelligibility and naturalness. It's really helpful to hear right away if the lyric "works."

I find that I write to the singer, and what works for that singer. The reality is that if someone else were to sing it, they'd "make it their own" anyway. Not that anyone is clamoring to sing any of my songs. :p

I've had to adjust keys, because AI or not, the limitations of the actual singer are there in the SynthV version. But that's not really a problem. It helps make sure that I don't keep the singer on excessively high notes, and instead use them more sparingly where they really count.

One of the things I like best about using SynthV is that it allows me to focus on the song, and not my voice and its limitations. The question changes from "Do I sound good?" to "Does the song sound good?"

And that detatchment and decoupling from my own vocal skill makes it a lot easier to go in and fix the melody when there are issues.
 
This is super interesting — especially the part about lyrics that technically fit but still sound “off”.

I’ve been running into exactly that problem:
even when syllables match, something in the rhythm or stress just feels wrong, and then you end up tweaking notes, lyrics, pronunciation… over and over.

Out of curiosity:
how long does it usually take you to get a line to feel “right”?

For me that iteration loop can get pretty time-consuming.
 
Out of curiosity:
how long does it usually take you to get a line to feel “right”?
Unfortunately, there are two parts to this: getting the line to scan, and getting the line to say the right thing.

Getting the line to scan is usually a matter of making sure the strong syllable (usually the first) lands on the strong beat. So for example, if you put "somewhere" on a weak beat, that's going to place the second syllable ("where") on a strong beat and sound wrong.

So - for me - badly placed stress is easy to hear. Fixing it can be as simply as making a prior note longer, pushing the stress on the following note to the next beat.

That make is a pretty mechanical exercise.

Now, getting the right words, that's a different problem altogether. :cry:
 
Yeah, that makes a lot of sense.

The stress/scan part is actually the easier one once you hear it — like you said, it becomes almost mechanical after a while.

What I’ve noticed though is that the real time sink is the combination of both:
you try to adjust the phrasing to fix the stress, which then changes how the line feels, which then makes you question the words again… and you end up going in circles.

That loop is what I find really hard to break.
 
Back
Top Bottom