Day 39 of #100daysofnetworks
Music Scale Similarity Network
Hi everybody. Thanks for your patience since my previous post. I had some personal life to deal with and needed some quiet. However, in the time that has passed, an explosion of music learning has happened with me, and I want to share about that today, as this blog and network analysis is responsible.
Help Me Help You (And Others)
Before I start, I have a favor to ask. I have written 38 of these articles already, as well as a book. Each one of these articles takes hours and hours of prep time, and I gave up every summer for over a year in order to write my book.
So, if you can, please do help this blog grow, if you find the content interesting. If you can just tell one of your friends about it, even in passing, it’ll grow. If you know someone who is interested in Data Science, Natural Language Processing, Machine Learning, Artificial Intelligence, Graph Theory and Network Science, or even music, send them my way!
I will be using social media less. I will not be posting on LinkedIn as frequently about this blog. If you can share my articles there and on other platforms, it also helps me.
Please, help me help others. I am not doing this for money or recognition, I am doing this to share useful knowledge and skills. I’m happy with the growth, but I’m not going to push it on social media as much anymore.
Scale Similarity Graph
I have wanted to do what I am showing today for a while, because I already knew it was possible. I have done similar things with text and also using other matrices (correlation), and the visualizations are really useful.
On Day 37, I showed how to create a Scale Similarity matrix and heatmap. This is great, but it is not as intuitive for use as a network visualization. Here is the heatmap, as a reminder.
That is usable, but it is a lot of information packed into a small amount of space. It is useful, though. For instance, if I want to find a scale that is similar Harmonic Minor Mode 3, I can look for the darkest square on the same line that is less than 1. Then my eyes have to look all the way to the side without losing the line and find the similar scales. That works, but what a pain.
Here’s the thing: you can build a graph using an adjacency matrix, and a similarity matrix will work, and so will a correlation matrix. But you have to set a threshold for it to work.
That’s fine. Let’s say that I want to see the network of scales that are 80+% similar. What’s that look like?
This is IMMEDIATELY useful to me as a musician, much more than a heatmap will ever be. What if I lower the similarity threshold a little, so that the rest of the network is mapped together? Here it is with the threshold set to 75%.
I have never seen anything like this as a musician, and it immediately made my learning more enjoyable. I no longer needed to just repeat scales and hope for memorization to happen. I can now SEE that there are what I call families of scales. The closer the nodes (scales) are together, the more similar they sound.
I can SEE what scales sound most similar. It’s like being able to see sound!
So, I used this new knowledge and split up all the scales I was learning into four groups.
After a lot of repetitious practice, I started to acquire systems thinking about scales. I could see them as families. I could practice them in sets of seven instead of more than twenty. Learning became fun and accessible, and memorable. It is easier to remember names when you have seven to learn from instead of dozens.
For each scale, I mapped out the notes from the eighth fret, C.
And with that, I practiced, practiced, practiced. I spent so many hours just experimenting, building skill and speed, learning names, and fusing together scales. This thing that always intimidated me became fun and memorable. Learning should be like that. Graph can help, but only if you use it.
And I can also compare scales programmatically. Yes, this is gnarly code, but I don’t care. It is not going to be used in production. It is quick and dirty to get information.
I can see that the difference between Harmonic Minor Mode 3 and Harmonic Major is:
Harmonic Minor Mode 3 uses a A-flat and Harmonic Major uses a G
Harmonic Minor Mode 3 uses an A and Harmonic Major uses an A-flat
But the rest of the notes are identical
And so on…
Music Learning Explosion
This is no exaggeration. Since Day 37, this Graph + Music combination has led to an explosion of music learning and inspiration for me. Other musicians have also told me that they are finding this useful (read day 37).
And since this weird, awkward, intimidating thing became fun, easy, and useful, the floodgates of learning really opened.
It has led me to:
Really loving my time practicing scales, and internalizing the knowledge
Getting familiar with the entire fretboard, not just repeating old stuff I know
Learning to use triads in arpeggios
Pulling my completely neglected 25 year old Mandolin out of the closet and learning how to play it a bit. I restored it, strung it, and now play it.
Really fun conversations with my fellow musicians, showing how Data Science could be useful in creating tools for learning in new ways
It’s Why I Wrote My Book
I didn’t write my book to teach you to be good at your job. I wrote my book so that you could learn to explore your own reality. I write to give you tools to enrich your own life. That is my why.
I do this because I know this is useful, and I know that if you learn this, you will be good at your job, and you will have a more enjoyable life, if you use it. Science is about exploring and understanding reality and the human experience. Graph analysis gives higher level understanding than staring at numbers or a heatmap. A graph visualization is sometimes exactly what you need.
Bear With Me
Please bear with me. I ALWAYS feel like sharing knowledge, but I DON’T ALWAYS feel like writing, and I am not always in the right mindset for writing. I don’t always feel like writing, but I do always feel like playing/practicing music. My music learning does really well during those times, but I am not able to write about it. So, a long time sometimes goes by, between articles.
And then when I do feel like writing about it, I don’t feel like promoting my writing on social media, taking screenshots and pasting them on various platforms, etc.
I love sharing information. If you enjoy the information I share, please help me help you by helping promote my material. I don’t make money for this. It isn’t my job. It is just something I enjoy, and I love helping people learn useful skills.
And I know that not everyone is a musician, so this blog has taken a bit of a turn. I told you in the beginning that there are many kinds of networks. Scale networks are one of them. Later, I’ll show you a cool way of exploring street networks. There is something for everyone in Graph and Network Science.
What Are You Going to Do?
I’ve now written 39 of these articles, showing many ways that Graph Analysis is useful. Hopefully, it has given you some ideas for using it in your own life and work. What are you up to? What are you using it for, or thinking about using it for? Have an idea and you need someone to hear you out? Do something with this!
In order to find usefulness, you have to start using this, and learning from the insights you discover. Let me know if you get stuck. I challenge you to start playing with graphs on your own a bit. Please read my old articles or my book if you want to get back to the basics.
That’s All for Today
Thanks to everyone who has been following along with this series. Happy learning! If you would like to learn more about networks and network analysis, please buy a copy of my book!