What you’re watching in the video above doesn’t involve cameras or motion sensors. It’s the kind of brain-to-machine, body-to-interaction interface most of us associate with science fiction. And while the technology has made the occasional appearance in unusual, niche commercial applications, it’s poised now to blow wide open for music – open as in free and open source.

Erasing the boundary between contracting a muscle in the bio-physical realm and producing electronic sound in the virtual realm is what Xth Sense is all about. Capturing biological data is all the rage these days, seen primarily in commercial form in products for fitness, but a growing trend in how we might make our computers accessories for our bodies as well as our minds. (Or is that the other way around?) This goes one step further: the biological becomes the interface.

Artist and teacher Marco Donnarumma took first prize with this project in the prestigious Guthman Musical Instrument Competition at Georgia Tech in the US. Born in Italy and based in Edinburgh, Scotland, Marco explains to us how the project works and why he took it up. It should whet your appetite as we await an open release for other musicians and tinkerers to try next month. (By the way, if you’re in the New York City area, Marco will be traveling to the US – a perfect chance to collaborate, meet, or set up a performance or workshop; shout if you’re interested.)

Hypo Chrysos live at Trendelenburg AV Festival, Gijon, Spain, December 2011.

CDM: Tell us a bit about yourself. You’re working across disciplines, so how do you describe what you do?

Marco: People would call me a media and sound artist. I would say what I love is performing, but at the same time, I’m really curious about things. So, most of the time I end up coding my software, developing devices and now even designing wearable tech. Since some years now I work only with free and open source tools and this is naturally reflected in what I do and how I do it. (Or at least I hope so!)

I just got back from Atlanta, US, where the Xth Sense (XS) was awarded the first prize in the Margaret Guthman New Musical Instrument, as what they named the “world’s most innovative new musical instrument.” [See announcement from Georgia Tech.]

It’s an encouraging achievement and I’m still buzzing, specially because the other 20 finalists all presented great ideas. Overall, it has been an inspiring event, and I warmly recommend musicians and inventors to participate next year. My final performance:

Make sure to use a proper soundsystem [when watching the videos]; most of the sound spectrum lives between 20-60Hz.

Music for Flesh II live at Georgia Tech Center for Music Technology, Atlanta, USA, February 2012. Photo courtesy the artist.

You’re clenching your muscles, and something is happening – can you tell us how this XS system works?

Marco: My definition of it goes like “a biophysical framework for musical performance and responsive milieux.” In other words, it is a technology that extends some intrinsic sonic capabilities of the human body through a computer system that senses the physical energy released by muscle tissues.

I started developing it in September 2011 at the SLE, the Sound Lab at the Edinburgh University, and got it ready to go in March 2011. It has evolved a lot in many ways ever since.

The Xth Sense wearable biosensors by Chris Scott.

The XS is composed of custom biophysical sensors and a custom software.

At the onset of a muscle contraction, energy is released in the form of acoustic sound. This is to say, similarly to the chord of a violin, each muscle tissue vibrates at specific frequencies and produces a sound (called Mechanomyographic signal, or MMG). It is not audible to human ear, but it is indeed a soundwave that resonates from the body.

The MMG data is quite different from locative data you can gather with accelerometers and the like; whereas the latter reports the consequence of a movement, the former directly represents the energy impulse that causes that movement. If you add to this a high sampling rate (up to 192.000Hz if your sound card supports it) and very low latency (measured at 2.3ms) you can see why the responsiveness of the XS can be highly expressive.

The XS sensors capture the low-frequency acoustic vibrations produced by a performer’s body and send them to the computer as an audio input. The XS software analyzes the MMG in order to extract the characteristics of the movements, such as dynamics of a single gesture, maximum amplitude of a series of gestures in time, etc.

These are fed to some algorithms that produce the control data (12 discrete and continuous variables for each sensor) to drive the sound processing of the original MMG.

Eventually, the system plays back both the raw muscle sounds (slightly transposed to become better audible, say about 50/60Hz) and the processed muscle sounds.

I like to term this model of performance biophysical music, in contrast with biomusic, which is based on the electrical impulses of muscles and brainwaves.

By differently contracting muscles (which has a different meaning than simply “moving”) one can create and sculpt musical material in real-time. One can design a specific gesture that produces a specific sonic result, what I call a sound-gesture. These can be composed in a score, or improvised, or also improvised on a more or less fixed score.

The XS software has also a sensing sequencing time-line: with a little machine learning (just implemented few days ago) the system understands when you’re still or moving, when you’re being fast or slow, and can use this data to change global parameters, functions or to play with the timing of events. For example, the computer can track your behaviour in time and wait for you to stop whatever you’re doing before switching to a different set of funcions.

The XS sensors are wearable devices, so the computer can be forgotten in a corner of the stage; the performer has complete freedom on stage, and the audience is not exposed to the technology, but rather to the expressivity of the performance. What I like most about the XS is that is a flexible and multi-modal instrument. One can use it to:

• capture and playback acoustic sounds of the body,
• control audio and video software on the computer, or
• capture body sounds and control them through the computer simultaneously.

This opens up an interesting perspective on the applications of the XS to musical performance, dance, theatre and interaction design. The XS can also be used only as a gestural controller, although I never use it exclusively this way. We have thousands of controllers out there.

Besides, I wanted the XS to be accessible, usable, hackable and redistributable. Unfortunately, the commercialized product dealing with biosignals are mostly not cheap and — most importantly — closed to the community. See the Emotiv products (US$299 Neuro Headset, not for developers), or the BioFlex (US$392.73). One could argue that the technology is complex, and that’s why those devices are expensive and closed. This could make sense, but who says we can’t produce new technologies that openly offer similar or new capabilities at a much lower cost?

The formal recognition of the XS as an innovative musical instrument and the growing effort of the community in producing DIY EEG, ECG and Biohacking devices are a clear statement in this sense. I find this movement encouraging and possibly indispensable nowadays, as the information technology industry is increasingly deploying biometric data for adverts and security systems. For the geeky ones there are some examples in a recent paper of mine for the 2012 CHI workshop on Liveness.

For those reasons, the XS hardware design has been implemented in the simplest form I could think of; the parts needed to build an XS sensor cost about £5 altogether and the schematics looks purposely dumb. The sensors can be worn on any parts of the body. I worked with dancers who wore them on the neck and legs, a colleague stuck one to his throat to capture the resonances of his voice, I use them on the arms or to capture the pumping of the blood flow and the heart rate.

The XS software is free, based in Pd, aka Pure Data, and comes with a proper, user-friendly Graphical User Interface (GUI) and its own library, which includes over one hundred objects with help files. It is developed on Linux, and it’s Mac OS X compatible; I’m not developing for Windows, but some people got it working there too. A big thumb up goes to our wonderful Pd Community; if I had not been reading and learning through the Pd mailing list for the past 5 years I would have never been able to code this stuff.

The Xth Sense software Graphical User Interface. Built in Pd.

The public release of the project will be in April. The source code, schematics, tutorials, will be freely available online, and there will be DIY kits for the lazier ones. I’m already collecting orders for the first batch of DIY kits, so if anybody is interested please, get in touch:
http://marcodonnarumma.com/contact

I do hope to see the system hacked and extended, especially because the sensors were initially built with the support of the folks at the Dorkbot ALBA/Edinburgh Hacklab. I’m also grateful to the community around me, friends, musicians, artists devs and researchers for contributing to the success of the project by giving feedback, inspiring and sharing (you know who you are!).

Thanks, Marco! We’ll be watching!

More on the Work

http://marcodonnarumma.com/works/xth-sense/
http://marcodonnarumma.com/works/music-for-flesh-ii/
http://res.marcodonnarumma.com/blog/

And the Edinburgh hack lab:
http://edinburghhacklab.com/

Biological Interfaces for Music

There isn’t space here to recount the various efforts to do this; Marco’s design to me is notable mainly in its simplicity and – hopefully, as we’ll see next month – accessibility to other users. I’ve seen a number of brain interfaces just in the past year, but perhaps someone with more experience on the topic would like to share; that could be a topic for another post.

Entirely unrelated to music, but here’s the oddest demo video I’ve seen of human-computer interfacing, which I happened to see today. (Well, unrelated to music until you come up with something this crazy. Go! I want to see your band playing with interactive animal ears.)

Scientific American’s blog tackles the question this week (bonus 80s sci-fi movie reference):
Brain-Machine Interfaces in Fact and Fiction

I’ve used up my Lawnmower Man reference quota for the month, so tune in in April.

Shimon, an adaptive, improvisational, percussion-playing robot, is getting smarter – and more famous, with appearances in places like the Stephen Colbert show. Now, humans have been known to get a big head under such circumstances. Shimon’s head has gotten “more social” – gestural intelligence helps the robot relate to fellow players and nod its head in time to the music.

I got a chance to talk more to project creator Dr. Gil Weinberg, director of the Georgia Tech Center for Music Technology. He’s also taken some of this technology and built it into mobile app ZOOZBeat which you can spot in one of the videos here alongside the (much more expensive, no you can’t have one) robot.

What might surprise you about the Shimon project is that it isn’t just about copying what humans can do with a robot. In fact, if anything, says its creator, it’s about human-robotic relations. “The project was always aimed at creating new and inspiring interactions between humans and robots, with the goals of creating new and exciting musical outcome that cannot be created otherwise,” says Weinberg. He emphasizes that, while the robot assimilates human listening, it has a musical style all its own.

And yes, if there’s any doubt that Georgia Tech students can kick our ass in smarts and drum chops, even the ones who aren’t advanced robots, that’s Caity in the new promo video at top, an architecture major with LEED certification and drum line experience right up to the pros (having drummed for the Atlanta Falcons). That’s their fight song at the end. In fact, the only thing Georgia Tech students can’t do, evidently, is sing. (Though, based on my undergrad alma mater, I really, really, really can’t make fun.)

Shimon on Colbert from Georgia Tech on Vimeo.

Talking Machine Music

CDM: You talk in the NPR interview about not wanting the robot to play just like a human. So, I wonder – would you think of the robot as an expression of the musical taste and instincts of its designers? In some ways, it does sound different from a human; could we think of Shimon as having its own style, or being interesting because of its non-humanness?

Weinberg: Our motto is “listen like a human, play like a machine”. The idea behind it is that in order to connect with humans, Shimon has to understand music the way we humans do. For this purpose we developed perceptual modules based on music perception research for concepts such as tension and release, stability, similarly, etc. When Shimons responds, however, we want him to be surprising and inspiring, introducing new ideas that humans are not likely to use, whether by using mathematical processes that humans cannot process in real time or just through mechanical abilities. For this purpose we developed algorithms that utilize concepts such as genetic algorithms, fractals, morphing of HMM-based improvisation, etc. So for example, Shimon can respond by morphing the styles of Monk, Coltrane and his human co-player, in a way that humans probably will never use. In that sense he has its own musical style. For each different piece, though, Shimon may have a different style or “taste” based on the algorithm we use. I assume that one could say that his “taste” is a combination of all of his styles, which are inspired by the designers’ input.

How has the robot’s algorithm evolved since we first talked about it? What sorts of modifications have you found useful?

An important recent addition to the project is Shimon’s social head, developed with the help of my former post doc – Guy Hoffman. In order to create the connection with humans, we explored ideas of embodiment and gestures, as instrumental aspects of expressive musical group play. Currently Shimon can detect to the beat and nod his head accordingly, which helps humans get into the groove. He would look at what he finds interesting (if one player is playing something different than before, or different than the other players, Shimon would look at her rather than the other players). If Shimon plays something sophisticated, he is more likely to look at his own arms etc. We are also working on anticipating and coordination. We installed a camera in his head, and are currently working on letting Shimon use the visual input to anticipate and coordinate his playing with humans, along with the auditory information it currently processes.

Did the Colbert Bump go to Shimon’s head?

First Shimon was somewhat insulted, for himself and for the idiom of Jazz in general. Then he realized that they spelled his name correctly, so he started bragging about it. He also started to play in other styles to distance himself from the genre

But seriously, I am wiring a new piece for Shimon in an African Marimba Band Play style, which will help show its versatility in genres. (And here is an old clip where he plays an Indian Raga.)

Any other work we should know about, or other research growing out of the Shimon project?

To validate the importance of visual cues in music group play (i.e, validating the importance of Shimon’s embodiment and physically of the robot in comparison to interacting with computer generated music) we conducted this research: Visual Cues:
The Effect of the Visual Modality on Musical Ensemble Synchronization

Also, check out this paper, which won the Best Cognitive Paper in ICRA 2010. [That's the IEEE International Conference on Robotics and Automation, for those of you not in the know. And IEEE originated as the Institute of Electrical and Electronics Engineers. Whew. -Ed.]
“Gesture-Based Human-Robot Jazz Improvisation” by Guy Hoffman and Gil Weinberg [PDF link]

Shimon in Videos

Georgia Tech Center for Music Technology

LaDiDa Demo from khush on Vimeo.

There’s no question iPhone/iPod touch development – really, just clever mobile development – has gotten a bit overhyped lately. But that’s all the more reason to do a round-up of genuinely interesting stories, real innovation happening on the platform. So, I’m clearing out my inbox with some of the more creative tools appearing recently on Apple’s mobile gadgets. There’s no better way to kick off today’s festivities than with this unusual “reverse karaoke” creation.

Sure, people may think they’re tone-deaf. But even the layperson has extraordinary powers of musical perception. So how could you train your iPhone to perceive and respond to music? That’s the question asked by LaDiDa for iPhone, the first of a new line of “intelligent” music applications for mobile devices. A “reverse karaoke” tool, the idea is to listen to singing and fake accompaniment, rather than having you sing along to canned backing tracks. Nothing is pre-programmed; everything is generated on the fly on the device.

It’ll even make up a Bollywood accompaniment to your singing:

LaDiDa Bollywood Duet from khush on Vimeo.

Of course, to me, it’s interesting not only what the iPhone is able to musically, but also what these algorithms are unable to make sound musical. Both reveal a whole lot about how we hear and conceptualize music. I think the team deserves real credit for making this fun, though, and on constrained hardware.

The app’s creator Khush follows in the footsteps of Smule in that it takes hard-core academic music research and uses mobile devices as a vessel for getting that tech in the hands (literally) of the general public. (See my interview with Smule founder and ChucK originator Dr. Ge Wang.)

Parag Chordia, developed at professor at the Georgia Institute of Technology and the gentleman you see in the video, spoke to CDM about what’s happening behind the scenes. He tells us about how this application was developed, and how the intelligent algorithms work (or at least try to work, as music analysis and auto-accompaniment remain at early stages).

First, an explanation of the app.

Khush CEO Prerna Gupta explains how it works:

1. You sing into the phone, and LaDiDa will compose music to match.
2. LaDiDa’s patent-pending technology analyzes the pitch and structure of the melody to compose a unique accompaniment for each recording.
3. To be clear, we do not query a database of pre-recorded songs. That is, LaDiDa has been designed to work with any music.
4. After recording your song, you can hear it with different styles. LaDiDa comes with three styles — E Piano Pop, Rhythm Synth Pop and Dub Tone — each of which has been developed using high-quality instrumentation to work specifically with our algorithm.
5. We will be launching new styles every month that will be made available through in-app purchases.
6. LaDiDa also works on rap! This month we’ll be adding three new rap styles.
7. After choosing your style, you can save the song and share it on Facebook, Twitter and email.
8. LaDiDa also has a Discover page, where you can hear songs recorded by other users from all over the world.
9. Khush was founded by music technology enthusiasts from the Georgia Tech Music Intelligence Lab. You can read about us here and also find out more about the research at our lab here.
10. LaDiDa went live in the iTunes store last week and is currently priced at $0.99.

Prena, the woman you see in the video, has some Web experience to boot, too, including founding a popular Indian dating site. Oh, and she’s a better singer than the music researcher, but, hey, that’s why we all went into computer music, right?

In case you’re wondering how you take a research idea and make it run on the iPhone – or how the algorithm works (and might get smarter in the future) – I turned to Parag for those details:

The initial code was developed in my lab in c++. Since the core algorithms are basically mathematical, that portion was relatively easy to port. However, we spent significant time thinking about how to optimize for the iphone and every aspect of the app, from the interface to sound design, has been built with the iphone in mind. For example, there are significant limits on sampler performance — samples have to be short and effects are more or less out — but we thought it was important for our styles to have a rich sound. So we
put great effort into designing light styles that sound realistic.

Another significant challenge was making the analysis robust to external noise; iphone recordings are lo-fi and corrupted with tons of background noises, which makes robust (and again computationally efficient) pitch detection essential.

Our approach to reverse karaoke is somewhat different than what’s been done before. A significant limitation of previous work was a lack of fine-grained key estimation, a problem that we felt was critical to successful vocal accompaniment (most people are not anywhere near a piano or an instrument with fixed tuning when singing into the app).

We also worked on trying to give some larger structure to the
accompaniment, which can often sound locally reasonable but notably lacking in direction. Again, a difficult problem particularly when people are singing snippets. Still it is sometimes possible to detect phrases, and we have tried to incorporate that information as well.

Auto-accompaniment is an endlessly fascinating and deep problem. As we learn more about human perception and cognition of music, as well as improve our tools for machine listening, our systems will become more musical. While we still have a ways to go, we believe that, with LaDiDa, we’ve created a product that is engaging and allows regular people to express themselves creatively.

If all of this talk about musical perception recalls the questions about how culture and background versus neurology can be used to explain music – as seen at the Notes & Neurons conference – that’s no coincidence. Researcher Parag played sarod with a fascinating ensemble at that same conference. Bobby McFerrin sings a really beautiful solo with the ensemble.

In fact, it’s absolutely worth contrasting the elegance and beauty of these all-human musical responses to the somewhat clumsy (sorry, Khush) iPhone responses. That’s not to say the iPhone creation is any less human – it’s a computation model programmed by humans, and is capable of some impressive feats made possible by their musical instincts and training. As such, we really can hear the gap between what advanced musicians can do intuitively and what we can model computationally, atop the restrictions of the device’s ability to sense the world around it.

World Science Festival 2009: Notes & Neurons, Part 5 of 5 from World Science Festival on Vimeo.