Ask the Founders: Q&A with Ryan Miyakawa

This month, we had our first-ever Q&A session, where Ryan answered questions submitted by the community.

We’re constantly blown away by the passion and enthusiasm of our community, and the response to this Q&A session was no exception. If we didn’t get to your questions this time, look out for future Q&A videos!

Watch the full video or read the transcript below to learn about the inner workings of Hooktheory and some exciting developments in the pipeline.

At what point in your musical career or theory studies did you realize you could make it “click” with this product? Thanks for all your hard work.

Thank you very much. The thing about Hooktheory is we never really intended for this initially to be a company. Chris, Dave, and I all went to university together, and at our school, we were essentially allowed to teach a class. They let students teach a class. They’d teach classes on the Simpsons, on Fortnite, whatever you want. We decided to teach a class on the theory of popular music.

So how did we come up with this idea? I’ve been a musician my whole life. Chris plays guitar, Dave played piano for many years, and I studied piano class. I was always kind of interested at an early age about what made my favorite songs tick.

I think, like many of you, when I hear a song that I really like, it evokes an emotional response, and I wanted to create that myself. At some point, we started thinking more critically about what are the patterns that people are using in popular music?

I should also mention that we all have engineering backgrounds, so that’s what we studied in school. So it’s kind of a more systematic, scientific way of approaching this problem.

But just after playing music for a long time and writing music, we realized that there are a lot of these patterns, and some of them, they teach you in kind of a conventional music theory, and some of them, they don’t. So we decided to compile all of our ideas, and we made this course, and we had about 40 students where we just kind of went over some of the basic concepts, some of the stuff that many of you’ll be familiar with from our books. At the end, the final project was everyone wrote a song, and we had a concert.

And it was really great because we were kind of engineers and had coding experience. In addition to writing a lot of the concepts in the book, we also developed some software to help create the examples in the book and to teach our courses.

This was really kind of like an early version of Hookpad. In the beginning, what we now call TheoryTabs, these were examples in the book. We created all these ourselves, so there wasn’t really a proper editor for doing that. And the purpose was really just to create examples to teach these concepts.

And once the course was over, it was really a lot of work. Any of you who are teachers know, it’s really a lot of work to teach a class. We decided we would just kind of put these materials online, and in the beginning, like nobody really used it. It wasn’t very good. The interface was super clunky. The early version of Hookpad was written in Flash, so it was doomed to be kind of a technology failure.

But I think the moment that it kind of clicked was when we decided to allow other users to start analyzing songs. At the time, the database was just basically the songs that we had analyzed for the examples in the book, but we were just really blown away by all the users who were super committed to analyzing music in the system. It was just super amazing. I mean, that was probably one of the best feelings that I had. You know, we had like maybe 200 songs that I had basically analyzed myself, and then in a couple of years, we had like 1,200 or 1,300. It was amazing to me that everyone wanted to be a part of this community. So that was really special.

And I think that’s when we started taking this a little more seriously and turned these things into products. This was really never supposed to be kind of a business at first. It was kind of more of a passion project. And I think that’s one of the reasons why we care so much about it now because it still feels like a passion project to us.

We all have day jobs in engineering fields, but we work very hard on Hooktheory in our off time because we care very deeply about it. And I’m not sure if this was supposed to be a business the whole time if we would feel the same way. So that’s kind of a long answer to that question, but that’s how we got our start.

Any plans on porting Hookpad to mobile (IoS/Android)? It’s been greatly helpful to sketch random ideas going through my head, I think it’d be really cool to be able to do so on the go.

So yes, we have thought about this a lot.

One of the big challenges with it is that it is very keyboard and mouse intensive, and it’s just one of those things that it’s definitely something that we eventually want to do. We need to figure out a way to not make it a horrible experience with your fingers. Because many of you probably have had some experience with mobile DAWs GarageBand on iPhone and an iPad. At least for me personally, the experience is quite a bit less productive than it is on my computer.

When we do this, we want to make sure that we have a plan, that this isn’t gonna be kind of a clunky, unusable product, that’s something that’s going be as useful to you guys as on the computer. So yes, this is definitely on our radar.

I understand sus2 and sus4. But I would like to know which one I should use and when?

So, the thing to understand about Sus2 and Sus4 is — this is true for any chord quality — it drastically depends on which chord you are sussing. So, say I’m in the key of C major. If you put a suss on a I chord, like the C chord here, that’s very different from the function of a suss chord on, say, a V chord.

So, tangibly what is the difference? What I would say is that on a I chord, it’s kind of an embellishment. So, for example, the song “Shut Up and Dance” is basically just a series of I chords with Sus4s and Sus2s.

Now, if you go to a V chord, typically speaking, the function of a suss chord on the V is to draw you very hard into the V because you’re setting up essentially a cadence back to I. Between Sus4 and Sus2, Sus4 is the one that has a stronger pullback to the third.

The reason for that is that the distance between the IV and the III, relatively speaking on the chord, is just a half step, whereas the distance between a II and a III is a whole step. So it’s typically understood that the Sus4 has a stronger pull to the third of the chord than the Sus2 does. So typically, when you’re setting up a pre-cadence, you’re going to want to use Sus4.

The short answer is, it kind of depends on the chord.

For a V chord, I would definitely recommend Sus4. I think for other chords like VI, I, probably either is fine.

On a IV chord, I’ll just mention real quick, you almost never want to use a Sus4 because the Sus4 for a IV chord gives you this flat seven, which is not in the key. So that’s going to, well, I shouldn’t say you should never do it, but understand if you are going to do it, then you are going to be creating non-scale tones in your songs. So typically, Sus2 is much more common on the IV chord.

What is the difference between a sus and an add? I know that add has the third and not the sus, but when would I use one versus another?

Okay, so this is a really good question. You’re correct in that add and sus are related. So in both cases, we’re essentially using some of the same skeleton.

So, let’s do I, Sus2, and then contrast this with an add9. What I’m going to do is I’m going to add the third now. So, an add9 has the third and the Sus2 does not have that third.

The thing about an add generally is, think of an add as, it’s like the vanilla chord, except you’re giving it a little more pizzazz.

So, like if I take a IV chord, for instance, the IV add9 is going to function exactly the same as the IV in most cases. What I mean by that is you can essentially swap the IV add9 for the IV, and you’re not going to change your chord progression much, but it’s going to kind of add this schmaltzy feel to it.

The suspension is different because you are intentionally withholding that third to set something else up. So, if I play a five with the Sus4, I’m essentially setting up the expectation of the V. So I’d say that the primary difference between a sus and an add is the sus, you are intentionally setting up an expectation, this tension, whereas in the add, you’re kind of sprucing up the chord.

I see that the company is coming up with new products like Chord Crush. But the Hookpad product doesn’t seem to have had “a major update” in terms of new features, rhythms, and sounds. When can we expect a new update?

So, I know it doesn’t seem like it, but we are constantly working on Hookpad. So, what specifically are we working on?

We are always working on new features. There is a long list of features provided to us basically by you guys.

I mean, the way that we kind of decide what to work on is by your feedback. So, your emails and things on the forum. I know that many of you post things on the forum, and you have lots of great ideas. We can’t get to all of your ideas all at once. I know that’s frustrating.

I want you all to know that we are definitely listening to you. And we talk about this all the time, about what things we’re going to focus on. The reality is, we can’t implement every idea all at once. We pick the things that we think are most important. Oftentimes, unfortunately, the bugs end up getting most of our focus because we feel really bad when something isn’t working.

So, for instance, what are we working on right now? We actually have a fairly significant update to Hookpad in the works. Dennis, an audio engineer that works with us, has created a host of new rhythms for drums and drum patches.

When we first introduced drums and Hookpad, basically, it was kind of like those little sequence editors you might see online where you’re like, “Okay, let’s put a kick drum here. Let’s put a snare drum here. And that sounds great.”

What we actually have now is Dennis, who is, by the way, a fantastic drummer. We’ve made it so that he can actually play his real drum kit, well, MIDI drum kit, and we can import those rhythms into Hookpad with all these little micro-timings. So the result is that the drums sound a lot more realistic, and it sounds great, actually. So I’m really excited to share that with you.

We also have some new synth instruments in the works. A lot, actually. A lot of new instrument patches. I apologize we haven’t released this sooner.

What tends to happen is that we start kind of working on these new features, but they get tied into these bugs that we’re also fixing. We kind of can’t release them until we feel like everything is working.

One bug that’s been plaguing us for a long time is there have been a lot of glitches with looping. A lot of you on the forum have been mentioning this, and so we’ve kind of reworked this whole thing. But I want to make sure this is all pretty solid before we get it out. So we’re working hard on getting that out. Hopefully, soon.

When are you going to start banning the trolls on your site purposely changing the keys of songs to be completely incorrect?

I really appreciate everyone’s patience on this issue. No one likes a troll. If you’re a troll, please stop trolling. It’s really a bummer when you work hard on something, and someone comes and vandalizes it.

So there are a couple of things that we have in the works. Obviously, we do our best to monitor these things. And we have, in the past, banned users on multiple occasions that are violating our terms of use or are just being kind of nasty.

I think one of the best things about our community is that everyone, generally speaking, is quite positive. And yeah, we’d like to keep it that way.

One thing that we’d like to do is to kind of implement a better way for people to let us know when things are off. So, one idea we had is to create flags for TheoryTabs, kind of like on Wikipedia articles. We can mark things that are even off about the TheoryTab. Maybe a video is broken, maybe something needs chords or a melody, or maybe this one is disputed, or maybe we can mark edits as being troll-y. That’ll generate a system for us to monitor this a little more closely.

But we’re also open to ideas that you guys have. Again, it’s definitely on our radar. It’s certainly something that we want to be addressing and just keep everything positive for the community.

If you could give one songwriting tip, what would it be?

There are obviously a lot of things we could talk about. Fundamentally, if I had to break down the art of creating a song, I would say that the most important aspect is managing tension.

So what do I mean by that? It’s like if you write a movie script or something that has no conflict in it, then it’s fine. But some of the best stories set up tension that then gets resolved later. Writing music is exactly the same way. There are many different ways that you can control tension, and I think that’s honestly one of the greatest things about writing music.

Some people are really into melody, some people are really into chords, and some people are really into mixing and instruments and production. Wherever your strength is, you can use that to control tension in some way.

For example, on the production side, you can control tension by giving more or less instruments, the frequency of beats, and whether you’re using drums or not.

In chords, which is something that I’m very passionate about, there are so many clever ways that you can control tension. One trend going back to the ads and sus question that we had earlier is the ad 11 chord. So we take a chord and instead of just having the 11 as the fourth scale degree, people are actually adding the fourth on top of the third to create this chord. On its own, this chord is so dissonant that nobody wants to hear it in context. But it’s crazy how nice it can be when you give dissonance some context.

One of my favorite ways to use this chord is on a V chord. So if we start in the I chord and I go to VI, on the V chord, you add this 11. It’s very dissonant on its own, but when it comes from the VI chord, it’s like our brains are more willing to hear this dissonance because this scale degree one is already very common to the key. So on its own, VI to V at 11 is palatable, but it’s clearly dissonant. We are creating this tension in the harmony. And then when we resolve it, a really nice way to resolve this is with the IV add9. It just feels really good and it feels better than if we didn’t do it right.

So if you just go VI to IV, that’s obviously a very classic progression, and people use that, and it’s fine. But it’s just so cool how you can add this dissonance and just make any progression a little more interesting. This is a very big topic, and maybe I’ll even try to expand on it in another video about the many different ways that you can control tension in chord progressions. But I think that you should be thinking about it because that’s like the knob that you want to be playing with.

One very easy way to get started is just by controlling tension between melody and harmony. We talk about this all the time in Hookpad. If you turn on the guides, we will show you which notes are in every chord. We call these stable notes. Stable scale degrees are the ones that are in the underlying chord, and unstable ones are not.

When you’re writing a melody, just try to mix it up. Do extreme things first. Write a melody that has only stable scale degrees and see how you like it there. Then write one that has only unstable scale degrees. Obviously, that sounds horrible. But start mixing them together and see how much tension you can create between the chords and the melody. Play around with that. If it’s too dissonant, you can always just use more chord tones.

I think that’s probably what expert songwriters do so masterfully. They create tension in melody in just really incredible ways and then just release it at all the perfect times. Sometime, we’ll talk more about examples of that, but for me, it’s just a joy to listen to and to analyze just a really professionally crafted melody and chord progression. So think about tension.

What are Hooktheory’s plans for better customer support?

So, I apologize if you feel like you need customer support and you don’t feel heard. If you need something from us, the fastest way to get a response is to email our support at support@hooktheory.com because we check that twice a day. If, for instance, you paid for Hookpad and you don’t have it, something like that, definitely email our support.

For feature requests and bugs, you’re welcome to email support too. You can also post on the forum. We don’t check the forum as frequently, but we do check it. So, we definitely appreciate your patience. We’re doing our best to get through all of these.

Hopefully, in the future, if we grow to the point where we can bring on a couple more people to help work on these bugs, that would be great. Right now, it’s just us. So, we’re doing our best. We definitely appreciate all of your support and patience, but we do try to get to all of the bugs.

Do you have any plans to introduce AI? Any pie in the sky things you think AI could enable music-wise?

Yes, we are definitely trying to do some AI stuff. So we have a couple of things actually in the works.

One AI model which was recently published by someone at Google Magenta who does a lot of research in the intersection of computer AI and arts, he basically built a model that can listen to a song and essentially transcribe it or analyze it in Hooktheory notation. So he has this model called Sheet Sage, you can look it up.

But we are trying to build essentially that into Hookpad so that when you all are creating TheoryTabs, you can start with this AI model that kind of gives you a baseline and maybe edit it afterwards. It doesn’t get everything 100% right, but it’s pretty cool and it’s pretty useful. And if anything, just to kind of get a skeleton for chords and melody before you’re analyzing TheoryTabs, I think that would be really cool to do.

So we have to work out how to actually serve that model off of one of our servers. But we’re working on it, and hopefully, we can give you a demo of that soon.

Also, it would be awesome if we had some sort of ChatGPT-like assistant. For those of you who are programmers, you might be familiar with GitHub Copilot. This is an AI that’s essentially like having a copilot sitting next to you when you’re coding, and it’s pretty amazing. It can generate code, and when we’re coding Hookpad, it’s something that we use all the time.

The thing about these models that you have to understand, like ChatGPT and Copilot, which is also built on a transformer architecture like ChatGPT is, they are trained on just tons and tons and tons of data. Just crazy amounts of data. And I think that’s kind of the secret sauce to making them from kind of a cute little pet project to something bigger. And even with all of the music data out there, I’m not personally convinced yet that we have something of that caliber yet.

But there’s certainly a lot of cool work being done in this space, and we are doing our best to keep a thumb on that pulse. And we would love to have some sort of Copilot in Hookpad that could help you come up with chord progressions and melodies. So definitely on our radar. Hopefully, we’ll have updates to that in the near future.

A few times I’ve run into a bug where I can’t move a note or have to unselect and select something again or recreate the bar. It has been easy to work around the issue but is there a way to report bugs?

I’m sorry about that. Sometimes we use Hookpad and things like this crop up too. Obviously, we appreciate any type of feedback, but if you really want to help us get to the bottom of these, the best thing for us is if you essentially have a way to reproduce it. So if you give us a list of steps, like, “Oh yeah, I open up Hookpad, and I write in these notes and something can’t select,” we can find those right away.

The ones that are really hard are the ones that are super intermittent. “I do this, and I do this, and sometimes it works, sometimes it doesn’t.” If we can recreate the bug on our own, it’s a lot easier to find it. I’m not saying we’re not interested in squashing all of the bugs, but we really do appreciate your help on this.

How about some sound add-on packs and better control over the output score/lead sheet?

In terms of sound add-on packs, we are constantly trying to add new sounds and new genres, new instruments, and new patterns to Hookpad. We have a lot of stuff on the roadmap. I personally find it very rewarding to work on new audio functionality in Hookpad.

This is something that we don’t anticipate ever stopping. The better we can make Hookpad in terms of the audio experience, the better it is for everybody. So that’s definitely in the works, and I hope that you can expect an update on this soon in terms of control over output score and lead sheets.

There have been many feature requests associated with our sheet music export, and we try to get to as many of those as we can. What I will say is that we typeset all of our scores with LilyPond. LilyPond is a free, open-source application that can essentially convert text in their format to these nicely engraved PDFs.

We have an option in Hookpad to output the LilyPond source. So if there’s something like a grace note or some sort of marking that you want to notate, it’s relatively straightforward to download the Lily file and make that change in the Lily source.

But if you have specific suggestions on features you’d like to see in sheet music export, please let us know. This is something that we are constantly working on to make the export and sheet music export experience a better one for all.