Today, I’m in London doing a hands-on workshop on visual metaphors for music, and covering various topics filed under “synesthesia” at Music Tech Fest. It seems appropriate, with the subject matter on the brain, to revisit the topic of visuals and music in a series of posts.

When you make hardware, with knobs and faders, you’re constrained by physical space – the amount of room on a circuit board, the radius of a knob cap, the size of your fingers. But before you get there, the first step is to sketch an idea. Imagine if you could do that with a marker and paper and ink.

SketchSynth is the latest attempt at a drawable set of controls, letting you turn an illustration on paper into something you can actually use to make music. It’s not the first – this dream of being able to make things come alive with nothing more than a magical pen is an old one – but the execution looks superb. There are two basic approaches to the idea: one is to use some sort of conductive ink to turn the drawing itself into a sensor, and the other is to point a camera at the drawing and calculate where a user makes contact with the drawing. SketchSynth opts for the computer vision approach, by way of OpenFrameworks and the old standby of free and open source vision, OpenCV. (Kyle McDonald’s ofxCV does the heavy lifting.)

Conventional vision fits the task well: faders and knobs respond as expected, even though they’re only ink on paper. While it’s a drawn interface, and could look like anything, the behavioral metaphors all come from hardware: there are sliders, momentary buttons, and pots. Place the tool in “edit” mode, and the computer analyzes what you’re drawing; in “play” mode, the camera tracks your finger as you manipulate what you’ve drawn. The project goes one step further than many that have come before, by overlaying a projection calibrated with your drawing for interactive visual feedback as well as sonic. Sound in this case is provided by Pd, but OpenSoundControl (OSC) messages let you connect to other musical (or visual) tools. See more in the making-of vid, at bottom.

Creator Billy Keyes is working with the right mentor, too, completing this as research for Golan Levin, who has long explored the relationship of drawing and musical gesture. His Sonic Wire Sculptor was a seminal creation in connecting drawings to sound, using a tablet to produce three-dimensional “wire” structures and corresponding sound synthesis. His Messe di Voce performance piece neatly reversed the relationship, using the voice as the input to animate drawings and illustrations. More of Levin’s work at flong.com. See also: composer Xenakis’ UPIC, which lives on today as IanniX, a tool getting a lot of development attention.

Keyes’ project, already getting lots of blog buzz, is notable for its practicality and immediacy; it seems a tool many others might use and build upon rather than a single piece of art.

It’s lovely to see projects, particularly academic projects, come to some form of completion and clarity. Speaking of completion and clarity, at some point a proper survey of drawn musical interfaces seems a must, but that will have to come another day. Where can I get a nice full English breakfast?

I’m late to the party, but hat tip in particular to Creative Applications

Project Page at golancourses.net; Linux source is promised soon

I’ve covered a number of these sorts of projects over recent years.

Paper, Drawing as Musical Controller: A Round-Up (including a number of paper examples)

Drawing Sound: Crazy Touch Interface Sound Experiments with Usine, PC (using only a screen – and thus producing a very different experience of drawing

Imaginary Instruments: Marker and Paper as Controller (this 2009 project is almost a direct analog to SketchSynth, minus the projection)

iPhones, Pencils: Hand-Drawn Music Interactions, Tokyo Subway Mobile Jam (this one isn’t quite the same as the others, using pencil and paper to design an interface for a screen – but it’s nice to recall that you can do that, as well)

Pen and Paper as Graphical, Digital Music Score (from earlier this year, a more open-ended design)

Video: Sonic Wire Sculptor, 2003:

“So,” you say, “I’ve got a … and I want to connect it to a … to make music. How do I do that?”

One strong answer to that question, if you’ve got a Mac, is junXion. Developed by the landmark audio research laboratory STEIM – a hotspot in Amsterdam that for years has been imagining new ways of making music by connecting things to other things – it got a big update recently.

It takes lots of the inputs you might imagine (joysticks, mice, touchscreens, MIDI, OpenSoundControl, audio, Arduino-powered hardware and all of its sensors, and video sensing) and connects it to a lot of the outputs you might imagine (using MIDI or OSC). You can set up rules in between the input and output to make that connection musically meaningful.

OSC input and output wasn’t entirely optimal in past versions; a total rewrite now makes it work with useful OSC sources like the iOS TouchOSC and Lemur apps. You get nifty new Actions, like remote mouse control. You can use a Nintendo Wii “Wiimote”‘s infrared-sesnsing capabilities and vibration support. If you’re using video, you can now support multiple “blobs.” And the whole app promises to run faster and look better, with more help tags in the UI, and added stability.

75 € for the full version. You need Mac OS X 10.5 or later, including the latest 10.7 Lion. (Upgrades for version 4 are free; Lite users can upgrade for 60 €.)

http://steim.org/product/junxion/

Of course, talking about this doesn’t really make much sense; it’s better to see it in action. We have a whole bunch of videos from the folks at STEIM showing features like Wii and joystick control and video sensing from a camera – plus a couple of fascinating demo/tutorials submitted by users.

Let’s watch, shall we?

Via https://vimeo.com/steim/videos

Far from the walls of STEIM, though, intrepid users have concocted their own demos. Here’s a look at controlling Reason with a Wiimote:

Here’s a live performance, also controlled by Wiimote, in the modular live environment AudioMulch. The creator writes:

A basic soundscape in AudioMulch controlled by two Wii remotes via JunXion IV.

Buttons in Wii Remotes control: start and stop buttons, presets of the main mixer, transient parameter of the granulator, frequency of the pulsecomb_1 (processing the drum), a junxion-timer controlling the volume of the granulator.

X-Y-Z accelerators control: 10 harmonics of a frequency generator, parameters of the rissettone

And yes, a camera can be a Theremin:

Got your own solution using junXion – or another tool? We’d love to hear about it.

See also two fine Mac-only tools:
Osculator [Much like junXion, supports nearly anything as an input, adds advanced OSC routing]
ControllerMate [not music-specific, but very powerful modular game input utility]

In fact, what’s largely missing is easy solutions on Windows and Linux, though you can roll your own with a free tool like Pd, which also supports HID, Arduino, video, and the like.

It started as some compelling demos or proof of concept, but it’s plenty real now: the tools for translating movement, gesture, and dance from the body to interactive music march forward. Empowered by Microsoft’s Kinect and an artist-friendly toolchain, even a single, clever developer can do a lot. Sound designer, music producer, and Max/MSP developer Chris Vik of Melbourne has been one of those busy early pioneers, with an incredible tool called Kinectar.

So, the tech is cool and shiny and impressive: what about the actual music? And, even more importantly, what if all the hand waving and moving about could be meaningful? That’s the next step. For his part, Chris is teaming up with a dancer and choreographer to combine his compositional ideas with someone who knows how to move. The Dubstep-y demos (all below) are impressive, true, but the early tests of the work with the choreographer are simply beautiful, and demonstrate that wobble bass isn’t the limit of what this can do. They also turn the arbitrary arm-waggling into a part of the art.

And as for you: the software’s alpha, but you can fire up your copy of software like Ableton Live and grab this software for Mac or Windows and try it yourself. So if you don’t like the results – be the gesture-controlled basslines too wobbly, be they not wobbly enough – you can put your music, and your movement, where your mouth is.

At top, Chris shows off an early test of the dance collab. (There’s more to come.) Below, a tutorial that shows how this works with Ableton. And read on for more from Chris on what the work with the dancer is about, and what the tool can do.

Chris writes:

Since April 2011 I’ve been working solidly with the Microsoft Kinect, developing my software, Kinectar, to enable its use as a MIDI controller for performing music live. I’ve done a number of performances around Australia since I started the project, however, it’s safe to say that, although I would consider myself an electronic musician, I’m certainly no dancer. Enter, Paul…

Dancer, Paul Walker and I have joined forces to bring the Kinect controlled music concept into the world of contemporary dance. Recently we obtained a residency at PACT theatre (centre for emerging artists), where we spent the week developing different ways of implementing my Kinect music control system in a dance context.

My system is developed in Max and uses OpenNI drivers, OSCeleton and Ableton Live.

via Chris’ blog

CDM will check back in with Chris soon, because:

I’ve got some more videos to release over the coming weeks from a range of my different Kinect music performance applications, including controlling/conducting the Melbourne Town Hall Organ and a 100+ speaker Kinect-controlled diffusion performance. I’ll keep you posted when they’re released!

More on the software:

Kinectar Performance Platform is a toolkit developed by music producer Chris Vik to allow the use of Microsoft’s Kinect motion tracking sensor in computer-based music. The software is designed for electronic musicians to expand the way they control their music in a futuristic and extremely expressive way, using only the waving of hands and a small amount of creativity. It can be used to control the simplest of parameters like a filter or LFO, play notes and chords on a sampler or synthesizer, or be programmed to control an entire live-set through nothing more than gesture.

Key Features:

Movement Tracking UI allows manipulation of the Kinect’s human tracking capabilities, displaying all relevant data extracted from the hands location in 3d-space

Instrument Builder lets the user build virtual ‘instruments’ by outputting MIDI notes in three modes:

• Static – Produces a single note value. Useful for drum triggers, turning on/off effects within a DAW or feed that trigger into Kinectar to switch between presets using your gesture
• Solo – Do sweeping solos by selecting from over 40 musical scale presets or click the notes on the UI to make your own
• Chord – Create a progression of up to 8 chords per preset to play live

Global Flags lets you turn on/off Kinectar’s instruments using a MIDI note sent from your DAW, external MIDI controller or Kinectar itself

MIDI Preset Control lets you switch between Kinectar’s presets and instruments using a single MIDI note

Value Editor enables many more MIDI/OSC outputs, for controlling device values

Visual Metronome popout window sits on top of all programs to make it easy to see if you’re in-time when the music gets messy

It’s labeled “rough alpha,” so don’t expect a finished tool here, but you can go download it and give it a try (or learn more about what’s possible):

http://kinectar.org/download

And now, the obligatory (but quite awesome, Chris) Dubstep demo videos:

The latest in a long tradition of transforming hand-drawn graphics into music (see, in particular, Iannis Xenakis and UPIC), we see a computer-vision-powered pen-and-paper music generator. Kovacs Balazs writes:

This is a manual sounddrawer. Doesn’t need any sensors, but a camera, paper, colored pens. Doesn’t need sensor glove or reactable as well.

What I love about this, though, is that the resulting sounds are utterly crazy, a big collision of notes and sound.

By the way, UPIC lives on here in a very advanced program descended from the original tool:
http://www.iannix.org/en/index.php

From credits: Magyar Eötvös Ösztöndíj Alapítvány, UCSB-MAT, CSALÁD

More:
http://soundsofpictures.blogspot.com/2012/01/17.html

Compare the complex model of what a computer can use to control sound and musical pattern in real-time to the visualization. You see knobs, you see faders that resemble mixers, you see grids, you see – bizarrely – representations of old piano rolls. The accumulated ephemera of old hardware, while useful, can be quickly overwhelmed by a complex musical creation, or visually can fail to show the musical ideas that form a larger piece. You can employ notation, derived originally from instructions for plainsong chant and scrawled for individual musicians – and quickly discover how inadequate it is for the language of sound shaping in the computer.

Or, you can enter a wild, three-dimensional world of exploded geometries, navigated with hand gestures.

Welcome to the sci fi-made-real universe of Portland-based Christian Bannister’s subcycle. Combining sophisticated, beautiful visualizations, elegant mode shifts that move from timbre to musical pattern, and two-dimensional and three-dimensional interactions, it’s a complete visualization and interface for live re-composition. A hand gesture can step from one musical section to another, or copy a pattern. Some familiar idioms are here: the grid of notes, a la piano roll, and the light-up array of buttons of the monome. But other ideas are exploded into spatial geometry, so that you can fly through a sound or make a sweeping rectangle or circle represent a filter.

Ingredients, coupling free and open source software with familiar, musician-friendly tools:

• Two projectors
• A monome
• Processing, the elegant and artist-savvy free software for visual code
• Ableton Live and Cycling ’74′s Max for Live, acting as the interactive glue with the sound world
• Drumaxx, Image-Line’s tasty physical-modeled drum synth
• Native Instruments Battery, the sampled drum engine
• Eclipse, the free IDE, for Java coding in this case
• Community Core Vision and reacTIVision (based on our previous info, at least), free and open source community-based projects for making the interfaces you see in movies happen in real life.

Another terrific video, which gets into generating a pattern:

Now, I could say more, but perhaps it’s best to watch the videos. Normally, when you see a demo video with 10 or 11 minutes on the timeline, you might tune out. Here, I predict you’ll be too busy trying to get your jaw off the floor to skip ahead in the timeline.

At the same time, to me this kind of visualization of music opens a very, very wide door to new audiovisual exploration. Christian’s eye-popping work is the result of countless decisions – which visualization to use, which sound to use, which interaction to devise, which combination of interfaces, of instruments – and, most importantly, what kind of music. Any one of those decisions represents a branch that could lead elsewhere. If I’m right – and I dearly hope I am – we’re seeing the first future echoes of a vast, expanding audiovisual universe yet unseen.

Previously:
Subcycle: Multitouch Sound Crunching with Gestures, 3D Waveforms

And lots more info on the blog for the project:
http://www.subcycle.org/

Perhaps you’ve seen the demo videos, as people do astounding things by moving their body around and using the Kinect camera to make music. Now, a set of Max for Live devices makes it reasonably easy to access your body as input inside Ableton Live.

FaceOSC Mapper

Pictured at top, this builds on Kinect superstar coder Kyle McDonald’s face-tracking tool and lets you use your face – position and even facial movements – to control Ableton Live parameters.

Kinect Camera

For use with the V-Module and vizzABLE systems, you can plug in one or more Kinect cameras, and get tilt, distance filtering (to remove backgrounds), depth maps, RGB and IR modes, and plug in your depth-sensing camera for more goodness.

Kinect – OSCeleton

The home run: look at skeletal tracking for extremely precise human control of parameters, as seen in the video. It only gives your left and right hand, but stay tuned for further developments. See also this example patch.

Example video shows how to “track hand positions and translate to volume and send levels.” Not impressed? Remember, it’s a proof of concept: you can assign to other parameters, practice your movements, change the musical content, and even modify the patch to make it work better.

It’s Electronic Body Music! (Sorry, couldn’t resist. Kids, ask your … parents jeez we’re getting old, aren’t we?)

Ed.: Hacking Web databases to search sounds, remixing tools to automatically create dub tunes, cameras to sequence and analyze images in new ways, Montréal hackers have been busy. Trevor Knight writes from the event with full coverage from Canada, latest outpost of this global music coding phenomenon:

Music Hack Day made its first appearance in Canada at the end of September, painting the event with a Montréal flavour, complete with bilingualism, Montréal-style bagels, and even an appearance of Stephen Harper in a hack. Over the Saturday-Sunday event, musicians, programmers, and hackers scramble to create any sort of new music project.

With the help of dozens of laptops, gallons of coffee, several APIs and staff from such companies as The Echonest, Soundcloud, and Grooveshark, the assembled hackers churned out and presented 24 hacks in 24 hours.

Bruno Angeles took home first prize for his hack, FaCeQuencer, which uses computer vision and a webcam to control a squencer/looper and at the same time, outfit the user appropriately to the style of music.

Free Assembly, the hack from Graham Boyes, took second prize and recreates a target sound using a database of samples. It uses The Echonest Remix API for the analysis and finds target sounds sourced using Freesound.org’s API. The power of this hack was clear when Graham demonstrated using a drum and bass track as the target sound and a recording of a dog playing in water as the sample.

With a heavy presence of students and researchers from the Centre for Interdisciplinary Research in Music Media Technology (CIRMMT) and McGill University, several of the hacks incorporated data-mining. For example, David Weigl, Hannah Robertson, and Andrew Hankinson created wuzhear, a venue-based database of historical concerts in Montreal from the Montreal Jazz Festival website and last.fm’s API, including set information and playable with a Grooveshark widget.

The Wub Machine 2.0, from Peter Sobot, automatically creates Dub or Electrohouse remixes of an audio sample, while The Beatbox Machine allows one to record beatboxing and return a drum sequence replaced with actual drum samples.

For a complete list of hacks, check out http://wiki.musichackday.org/index.php?title=Montreal_2011_Hacks

Now that Canada has tasted the sweet Music Hack Day nectar, there’s already buzz for a hack day in Toronto.

Dac Chartrand (Renoise) writes with more details. Dac has his own set of hacks, but I’m excited enough about it that I’ll put that in a separate post -PK:

Other "Renoisers" were on site, but used the 24 hour session to try new ideas, but not present them. For example Steve Sinclair (Radarsat1) tried to port Mark Zadel’s Different Strokes to Android. Different Strokes resembles a freehand drawing application. The drawn strokes create animated figures whose motion is mapped to sample playback. The musician performs by assembling networks of strokes live, generating audio patterns. Steve got drawing and particles working but not enough time to hook into the Android audio subsystem, so he didn’t bother presenting.

Longtime CDM readers Studioimaginaire were also on site hacking away at their multi-touch table. They were there the full two days but also didn’t bother presenting. I tried to talk them into it several times, saying that the crowd would obviously vote for them just on cool factor alone and that they would walk away with a prize, but they stuck to their principles. Something to be said about the vibe of the event. Hackers were there to have fun.

David Viens of Plogue made am appearance for the demo session Sunday afternoon.

Good times had by all.

You might imagine sound in space, or dream up gestures that traverse unexplored sonic territory. But actually building it is another matter. Kinect – following a long line of computer vision applications and spatial sensors – lets movement and gestures produce sound. The challenge of such instruments has long been that learning to play them is tough without tactile feedback. Thereminists learn their instrument through a the extremely-precise sensing of their instrument and sonic feedback.

In AHNE (Audio-Haptic Navigation Environment), sonic feedback is essential, but so, too, is feel. Haptic vibration lets you know as you approach sounds — essential, as they’re invisible. The work of Finland-based DJ/VJ Matti Niinimäki, aka MÅNSTERI (“Mons-te-ri”), the project is part of research undertaken at SOPI Research Group at Media Lab Helsinki. Like some sort of sound sorcerer, the user is entirely dependent on movement, feel, and sound as they move unseen sound sources through space. (More technical details below.)

It’s labeled, as always, “proof of concept.” The creator promises more videos to come; we’ll be watching as this evolves, as it looks terribly promising.

Below, “Tension” is a fair bit simpler, in which users walk through a space and control synth parameters. (“You are the knob,” one might say, though I don’t suggest shouting that at someone you don’t know. They could take it the wrong way.)

More descriptions:

AHNE

This is a demonstration video of AHNE – Audio-Haptic Navigation Environment.

It is an audio-haptic user interface that allows the user to locate and manipulate sound objects in 3d space with the help of audio-haptic feedback.

The user is tracked with a Kinect sensor using the OpenNI framework and OSCeleton (github.com/​Sensebloom/​OSCeleton).

The user wears a glove that is embedded with sensors and a small vibration motor for the haptic feedback.

This is just the first proof-of-concept demo. More videos coming soon.

HEI Project 2011
SOPI Research Group
sopi.media.taik.fi/

Aalto University School of Art and Design

AHNE – Sound and Physical Interaction

Tension

A brief video showing Tension. An interactive spatial sound installation for multiple users.

A person enters the space and a generative sound is assigned to that person. The sound pans around in the 6-channel speaker system following the user in the space.

Up to 5 users can use the installation at the same time. Each person modifies the other sounds based on the distance to the other users. The closer you are to other people the more the tension in the sound increases.

Tension – Sound and Physical Interaction

Side note: watching these two videos makes me want to consult with someone on non-verbal expression, posture, and stage presence. That criticism is mounted at myself – I could use it. Perhaps we need an all-physical, unplugged music event for laptopists, controllerists, and electronic musicians. And I can at least say I’ve had some experience in this, working in the dance program at my undergraduate alma mater, Sarah Lawrence. Anyone game? (Sounds like something we could do while CDM is in Berlin in the fall.)

For their part, the Finnish research facility is working with dancers, along with Nokia Research Center. (Sadly, I can’t find documentation.) But I think interesting things happen when us non-dancers learn movement technique, too.

While we’re on the subject of making music with Kinect, the 3D computer vision camera with depth-sensing, here are some other experiments into how music might work. As with the classic Theremin, those musical gestures tend to be mapped against two-dimensional axes in space.

And from there, things become wide open. Johannes Kreidler, a musician and artist known for irreverent and inventive experiments in music, shares his studies for the Kinect, which he terms “conceptual music.” A solo “for violin” can involve literally waving a violin around. “House music” can mean making music whilst ironing a shirt. Any gesture in space becomes musical. Without tangible feedback, that can be challenging, and since these are just gestures in air, precision and nuance may not be a strong suit. But it’s a fascinating look into what’s possible, a set of thought experiments in music with a camera.

Composer Johannes Kreidler’s other works have included provocative ideas like making a performance of a short piece with 70,200 quotations of other “sampled” works, tunes from wildly-gyrating stock market quotes, entire bodies of work (like the Beatles) compressed into seconds, and pieces from avant-garde happenings to more conventional electro-acoustic scores. See his site for more.

And he’s got a killer book on using Pd, too.

In a whimsical proof of concept, artist and inventor Dan Wilcox harnesses the depth-sensing powers of the Kinect camera to turn a room full of drifting balloons into music. It occurs to me that the basic spatial model can be seen as descended directly from the Theremin – way to go, Leon, still relevant today. The sounds are simple, but it seems something you could continue to develop musically – to say nothing of what it could do for the ball pit at Chuck E. Cheese’s. (Slogan: Where a Kid Can Be a Kid Who Gets Obsessed With Skeeball Prizes / Get Scared Out of a Kid’s Mind By the Other Kids in the Ball Pit. Sorry, it’s an American suburban thing of a certain age, for the more than half of you who have no idea what I’m talking about.)

Full description from Dan:

Experiments in balloon motion and sound using an MS Kinect depth sensing camera.

Created for the Carnegie Mellon 1st & 2nd year MFA Graduate show entitled “Fresh Baked Goods” at Bakery Square, April 2011.

A machine stands in a room surrounded by balloons. Circulating fans blow the balloons over the machine which creates sound based on their movements.

Mode 1: Tones

Balloon height and x/y position control the pitch and panning of a treble and bass voice. The tones can be quantized into a certain key or a glisssando can be employed for a theremin-style effect.

Mode 2: 99 Luftballons

The playback speed of Nena’s 99 Luftballons is controlled by balloon height. The balloons must be kept in the air for the song to play. Feed the machine.

Built using Open Frameworks, ofxKinect, and Open CV for balloon tracking and Pure Data for sound generation/playback.

See danomatika.com/​blog/​balloon-project for more info.

Dan has a master plan with a robotic music-playing suit and other ideas, so I can’t wait to see where this goes.