Q+A: London musician creates AI tool to help write songs, lyrics

Article content

Western University music school graduate Jeff Lupker has developed an artificial intelligence tool, dubbed Staccato AI, that generates lyrics and melodies. It’s hit the right notes with Silicon Valley investors, has partnered with music label AMG and has a motto that sums up the goal: “Think of it as an AI Lennon to your McCartney.” Lupker, a 33-year-old Londoner, spoke with reporter Beatriz Baleeiro about incorporating technology into music-making and AI-related copyright concerns.

Article content

Q: What does the platform do, and for whom is made?

A: The idea is to make it easy for musicians, but we’re also thinking of what a creator means in a different way. I think how people can create music is sort of changing. So, we also have a version that works for more content creators who have music in their heads but don’t always have the means of communicating those ideas to a guitar or a piano or using digital audio software that’s available.

Q: Why did you create it?

A: I come from music. I started playing guitar when I was 13 years old. My whole background was in music first, followed by AI encoding and programming, but I’ve always been interested in how we could push music forward.

Q: Is there something special about Staccato compared to other existing AI-led content generators?

A: We’re trying to differentiate ourselves from other AI music tools. There’s some out there where you can write a little bit of text, and it’ll give you a full, fleshed-out song. Ultimately it’s the humans that are going to be the ones that push arts and music forward. So, we’re trying to give them all the tools so that they can sort of succeed and be better kind of music creators. They’re the ones that are going to have to put the finishing touches on it.

Article content

Article content

Q: How does it work?

A: For those who like writing with instruments, you can record a piano part into it, and it’ll continue the music right there. You don’t have to tell it the keys, the beats per minute, or anything like that. It will just sort of figure out all that information based on what you gave it. For people who know how to use some of these tools but still don’t know how to record the stuff themselves, they can write in text prompts.

Q: Do you think musicians would admit to using it? That seems like a pretty big hurdle to the company’s success.

A: We’re just another tool out there that happens to use AI. So, whether you already have the ideas or you don’t, you just need a place to start. It’s like you’re working with this tool rather than relying on it to do everything for you. We’re just a new system using some of the most current technologies, but ultimately, it’s just how you use it. We want to keep the creator in a constant state of creative flow, where they never feel like they’re hitting writer’s block. At times, it’s almost like a learning tool, and you’re just bouncing ideas off of it.

Article content

Q: The platform produces melodies and lyrics. What does this mean for the musician’s creative process?

A: I’d be much more reluctant to take my own lyrics and show my friends. If I were working with this tool for whatever reason, it would give me confidence that something else validates what I’m working on. If you’ve absolutely nothing and you’re a beginner, you might rely on it quite a bit. We’re going to give musical building blocks such as lyrics, chords, and melodies, but you’ll have to stitch them all together to make something interesting there.

Q: You come from a musical background. How do you feel about potential copyright issues?

A: I’m looking to make something for musicians, and the idea of using copyrighted material doesn’t really make sense for us as a company. We want the best musicians to use it, so we want them comfortable that their stuff isn’t going to be suggested to somebody halfway around the world. We don’t want anybody feeling like that, even just from the start.

Q: What is the company doing to avoid copyright problems?

A: We’re working on this checker that can take any section of the song created on our platform and can check it against our database of other stuff that people have used. It can tell you when there’s a melody that’s too close to something else. We’re also trying to work with our lawyers because we don’t suggest the same chords. We want to make sure we capture melodies so that people can feel comfortable using our tools. Someone could still take it off our platform and just completely rip off a melody or add extra stuff and publish it. At that point, they’re really going to get in trouble themselves because you can’t steal somebody else’s music in general.

Article content

Q: How are you preventing these issues?

A: Our underlying technology isn’t all audio files. So, we use kinds of representations of music. Largely for what we do, it’s going to be mostly melodies that are going to be the cause of problems for anybody. If someone’s working on a melody and they’re about to export it from our site, it doesn’t mean that it’s going to go straight to Spotify or anything like that. If people try to take it off our platform, we’re going to do everything in our power to make sure that we’re stopping something that has the potential to be too close to something else.

Q: How does a musician release the music created on the platform?

A: We don’t generate that final, polished audio file; you’re more in this pre-publication standpoint. We’re more at that compositional level where you could create the song, but then you would have to go and release it yourself and publish it or take it to a publisher.

Q: How is it being part of the Music Canada advisory council in times of AI-related copyright issues, and what’s next for Staccato?

A: Part of it is chatting with the people there in Ottawa involved in these decisions. We (Staccato) just make sure that we’re kind of doing Canada proud on this kind of international scale of how we’re going to handle, copyright situations.

bbaleeiro@postmedia.com
@BeaBaleeiro

Article content