Natural Machines


The Natural Machines album is out today, Saturday October 27th. I’m bringing the project to Le Poisson Rouge in New York on October 30th. The fine folks there asked me some smart questions ahead of the gig. Thanks to John Ruscher for asking me to do this interview.


LPR: Can you summarize what Natural Machines is?

Dan Tepfer: It’s a project where I explore the intersection, in music, between natural and mechanical processes. I do free improvisation at the piano, and programs I’ve written on my computer interact with me in real time as I’m playing, both musically and visually. I love it because it feels like very new territory, and it’s exciting to be exploring it.

You’ve mentioned you had a background in computer programming and physics as well as music. Does a relationship exist between the hard sciences of computer science and physics and the art form of music, and if so, how would you describe that relationship?

All great music, and probably a fair amount of great art too, but definitely this is true of music — all of it exists at the intersection of the spiritual and the algorithmic, I think. Good music, like good architecture, has strong structure. It’s what keeps it from falling apart. So whereas the surface of music is often above all emotional / spiritual, in the best music this surface is supported by logical systems, of varying kinds; this is what gives the surface a feeling of depth and stability. Just as an example, I don’t think it’d be that far fetched to say the “art” of counterpoint is close to a hard science, in the sense that it offers quite definite right or wrong answers to certain musical problems, and counterpoint is the underpinning of much of the music of the last 500 years, so there’s no doubt there’s a relationship there.

What is it like to play with an automated instrument such as your piano? Do you find yourself reacting as if you were playing alongside a real person, or is it a different kind of feeling?

It’s a different kind of feeling, because the computer responds according to strict rules I’ve programmed into it, so it would be like playing with a person that has absolutely no feelings but, instead, an unbelievably virtuosic ability to calculate. When I play and the algorithms respond, it feels more like I’m playing alongside an idea than a person.

You cited the music of Bach as being an example of that combination. Where do you see both the spiritual and the algorithmic in Bach’s music? Has it affected how you play Bach, and how?

Bach is really the poster child for this idea. His music speaks so strongly to us at an emotional and intuitive level, and yet when you dig into it even a little, you realize that it’s all supported by a mastery of logical systems. Often his music has more than one algorithmic process behind it. For example, in every third variation of the Goldberg Variations, he’s using, on the one hand, the rules of counterpoint, which by his time had been pretty strongly codified, and on the other, he’s imposing on himself the rules of canonic writing, where one voice imitates the other a certain musical interval apart. What makes it algorithmic is that Bach never cheats on either one — once he’s set up rules for himself, they are as rigid as if they were being calculated by a computer, with very few exceptions. And those aren’t the only rules at play. He’s also following a set chord progression, the same one in each variation, as well as adhering to larger structural ideas that govern the whole work. It’s no coincidence, for example, that the theme of the Goldbergs has 32 bars, divided in two at the midpoint, and that the work as a whole has 32 movements, likewise divided in two at the midpoint. And yes, getting deeper and deeper into the analysis side of Bach has deepened my playing of his music, no doubt about it. And it’s definitely given the act of improvising off of his music, as I do in my Goldberg Variations / Variations project, greater substance. 

In this performance, you’re programming an object (the piano) to act based on algorithms you create, and then recording how that object behaves, which seems very much like a science project. But, you’re also performing and communicating with that object inside an art form, which is music. Is Natural Machines a science project, an art project, or something entirely different?

Well, that’s the thing: I firmly believe that’s a false dichotomy — great art is naturally both at once, not one or the other. No one who’s really studied Bach would deny the science aspect of his work. But that’s not true only of Bach; it’s also true of a composer like Ligeti, or Bartok, and definitely of John Coltrane and Steve Coleman, too. 

It’s easy to talk about the technological aspects of the Natural Machines project, because technology is something that naturally lends itself to being put into words, but my most important focus in this project is making sure that the music is strong on its own terms. I’ve spent 11 years playing with Lee Konitz, I’ve spent my life performing Bach and playing jazz standards, so there’s no way that I would be happy presenting something that is only a science project. However, if I can find something that is both at once, and strongly so — that’s something I can get excited about.

Does the performance incorporate improvisation at all, and what kind of improvisation?

The project is very much based around the idea of free improvisation, and specifically, how free improvisation butts up against constraints. It’s a little like pouring water into a receptacle — whether it likes it or not, the water, even though it’s free flowing, will take the shape of the receptacle. So what I’m doing is creating a world, with its own natural rules of physics, and then visiting that world as a free agent and trying to find a way to enjoy living in it, to dance. I’m improvising from scratch, but the computer’s response makes it so certain things work, and others don’t. It’s the vessel I pour my freedom into. It shapes my freedom.

By now, you’ve probably spent a lot of time making music with this specially programmed piano. Because you’ve spent so much time with it, are you ever able to predict what the piano is going to do, and are there some patterns that the piano tends to play more than others?

The piano doesn’t really play patterns unless I’m playing patterns myself, because its reaction is always some kind of transformation of what I play. So, if I find myself not being surprised, it’s really up to me, as a free improviser, to not be complacent, and immediately search for a fresh angle to explore. But this is true of any situation in which you’re improvising, whether or not there’s technology involved. Only the great masters, Lee Konitz being a great inspiration in this regard, manage to keep it fresh every single time. It’s really important to me, in this project especially, that I keep surprising myself, and I’ve found that there’s nothing like painting yourself into a corner to make that happen. And there’s nothing like a computer to paint you into a corner…

What was your reasoning behind incorporating virtual reality into the performance?

Over the last year and a half, I’ve been creating visualizations for the different musical processes, with the intent of revealing the underlying musical structure of each. At shows, I’ve been projecting them on a screen behind me as I play.  Some of them are three-dimensional worlds, with a strong feeling of depth. They feel like a space that it’d be fun to explore, move around in, so I’ve been thinking of trying to incorporate VR for some time. But there’s a big challenge as far as incorporating VR into this project is concerned, because everything I’m doing is being created in real time. None of the notes I’m going to play are known ahead of time, nor the computer’s response, and hence none of the visuals either — only the way these things work is known ahead of time. So the data needs to be sent to each audience member’s device right as I’m playing it at the piano, and right as the computer is responding, with as minimal lag as possible. I think I’ve finally worked it out — I programmed it so it’ll all happen in a smartphone’s browser. I just ordered $600 worth of Google Cardboard, one for every audience member, and bought a powerful wifi router that should be up to the task of sending data to everyone at once. It’s going to be really exciting to see if it works at LPR, where I’ll be trying it for the very first time on a couple of pieces.

Do you think that in the future we’ll see more and more interaction between the arts and the hard sciences? If so, what kind of interaction? 

Art reflects the times we live in, and the times we live in, for better or for worse, are increasingly technological. So art needs to deal with technology at some level, whether it be by incorporating it, or rejecting it, or some middle ground. Music has always had a close relationship to technology. The pipe organ was one of the most sophisticated machines of its time when it was being developed, and the same could be said of the piano. Now, in terms specifically of hard sciences, Spotify now has an entire research department devoted to applying A.I. to music (imagine the profits once you have a program that can crank out infinite numbers of royalty-free tracks that sound good enough for people to put them on as background music! This has already happened, by the way). Also, while I’m on the topic, the mathematical building blocks of musical harmony were laid down by Pythagoras, so there’s a long history of the hard sciences interacting with music, and I can only see that increasing as we go. 

However, and this is a big but, as I said earlier, it’s not art if it’s just science. Art has to, somehow, speak from the soul, it has to tell us something unutterable about the mysteries of existence, or it’s no good. So the more scientific it gets, the more, if we want to keep making art that makes a difference in the world, we’re going to have to dig deep within ourselves to satisfy the left side of the spiritual / algorithmic divide.

What do you hope the audience will gain from Natural Machines?

Most of all I hope it takes them on a journey. I want my music, and my shows, to be transportive — to pick you up, take you on a ride and gently put you down in a different place, hopefully changed. 

What do you think about potential applications for virtual reality in the arts? Do you have any future plans to incorporate virtual reality into your music making?  

Well, I’m excited to start exploring the possibilities at LPR. If it works, I’ll go from there. My default attitude towards new technology is to be skeptical, but to try it and see. When it really does make a difference, as I think what I’m doing with the piano and the visuals does, it feels really special, because in many ways you’re lucky if you find a way to incorporate technology into art in a way that genuinely transcends the gimmick. 

Do you think that it’s possible to use science to understand the arts better, and vice versa? If so, how can we go about that process? What are some ways that science has helped you understand music as an art and express yourself through it?

I think we can use science to understand the science side of music better, for sure. To me it’s fascinating, for example, to understand the whole-number vibrational math behind just intonation, the ancient mathematical method for making sounds that work harmoniously together. But science can’t help us understand the spiritual side of music, no sir. For that, we just need to live our lives, feel things as deeply as we can, listen hard, and when we open our mouths, try to make sounds that matter.

This entry was posted in Music. Bookmark the permalink.

13 Responses to Natural Machines

  1. Tim says:

    It would be great to know the tools and techniques you used to create Natural Machines.

    • dantepfer says:

      I wrote the code for the musical part of Natural Machines in SuperCollider, and the visual part in Processing. The new VR stuff I’m coding in Unity.

  2. Zac Boden says:

    You’ve done a great job implementing a very interesting idea! I’m interested to know how quickly the piano can respond? For example, what’s the fastest that your input can go through the algorithmic loop and generate a mechanical tone? Do you manipulate that delay in your algorithms? Similarly, are the algorithms set to specific tempo’s?

    • dantepfer says:

      Hi Zac,
      I do manipulate the delay in some of the algorithms. In the album version of Inversion, for example, I double the delay about halfway through. And in the album version of Canon at the Octave / MetricMod, the delay is constantly going through metric modulations, either being multiplied by 2/3 or 3/2. I make all these changes with foot pedals.
      As for the minimum delay, the computer calculation is essential instantaneous. The piano, because it’s a mechanical object, can’t act instantaneously, but using its fastest mode, it can be as quick as about 60ms for a loud note. Softer notes take longer, of course, because the hammer has to move more slowly towards the string.

  3. Pingback: TD Ottawa Jazz Festival – Dan Tepfer – Natural Machines – Wolf Banter Blog

  4. Pingback: Hear Bach's Music Like You've Never Heard It Before: Upside Down -

  5. Pingback: Hear Bach's Music Like You've Never Heard It Before: Upside Down | Biz Instant

  6. Pingback: Hear Bach's Music Like You've Never Heard It Before: Upside Down - News Azi

  7. Pingback: Hear Bach's Music Like You've Never Heard It Before: Upside Down - Techbyn

  8. Pingback: Hear Bach’s Music Like You’ve Never Heard It Before: Upside Down | Headline Booking Group

  9. Pingback: Hear Bach’s Music Like You’ve Never Heard It Before: Upside Down | WSKGWSKG

  10. Pingback: Dan Tepfer Performs Bach Like You have Never Listened to It Before: Upside Down : NPR | marthafied

  11. Pingback: Dan Tepfer Plays Bach Like You’ve Never Heard It Before: Upside Down : NPR - Myfestival Status

Leave a Reply

Your email address will not be published. Required fields are marked *