Music lovers are hacking Apple’s Siri voice recognition technology. By connecting to some of the “intelligence” of the cloud, these tools can make your phone rap or send music files to a player piano for instant musical playback.

First up: Yamaha’s piano taking requests, thanks to music grabbed online. A Yamaha rep explains:

Yamaha consultant Craig Knudsen demonstrates a unique implementation of Apple’s incredible Airplay technology in an exciting new way.
Here’s how it works:
Take a standard MIDI songfile and convert it to an audio file (while maintaining the MIDI data). The songfile is then sent wirelessly via WiFi to an Apple Airport Express (which is mounted underneath a Yamaha Disklavier reproducing piano. The audio output of the Airport Express is then connected to the analog MIDI inputs of the Disklavier, using a standard audio cable.
Then, you simply ask Siri to play your favorite song from your iTunes library, and Siri responds immediately, by making the Disklavier’s keys and pedal move up and down, recreating the performance, including full orchestration.
The result is nothing short of magical.

Of course, the actual “playing” is thanks to the capabilities of the Yamaha Disklavier. I’m actually a bit puzzled as to how the online conversion works, exactly, and I was curious for any Disklavier-owning CDM readers whether this is something publicly available. I’m waiting to hear back from Yamaha.

And now, for something completely different: Siri rapping. (Somewhat … erm … badly, if amusingly. It is a hack.)

My friend Robert “Robb” Böhnke had a lot of fun combining Siri’s voice synthesis and the lyrics of Notorious B.I.G.

My hack for the Music Hack Day 2011 in London, a 24h Hackathon for all things music.

SiriProxy is used to intercept the communication with Apple’s servers. Notorious Siri then sends Notorious B.I.G.’s Hypnotize to the device.

Siri’s speech synthesis is synced to the beat using the timestamps obtained from the Echonest API which were then manually tweaked, to smooth out delays in the text-to-speech engine.

Thanks to Universal Music for awarding me a nice pair of Dr. Dre headphones

The Echonest API, by the way, is an amazing do-everything “API for music,” one that analyzes musical files and connects to a vast storehouse of musical intelligence.

Warning: this video is most definitely Not Safe For Work. (Heck, even the thumbnail isn’t, exactly.) If that concerns you, just go listen to the Yamaha video again, okay?

Siri FAQ [Apple.com]

Previously: Google Translate Beatboxing

…and, of course, that means: Android, your move.

Digital or analog, the essence of recording and production is the act of representing. One thing becomes another; one medium stores information about another. That representation can also be physical, tangible, and visible. In a sculptural pianola, Akko Goldenbeld turns the Dutch city of Eindhoven into a pianola roll, so that the landscape of buildings and streets acts as a physical musical score. I think it raises some questions about whether translating the one into the other obscures the experience of a city rather than clarifies it, but that would discount the act of watching it: with the visual connected to the sound, one begins to see the topography of the resulting music.

From the description on the video:

The tall buildings in the city centre have a heavy touch; the low-rise villas to the South create considerably gentler sounds. Akko Goldenbeld has a very personal way of looking at, or rather listening to the city.

He has created a scale model of Eindhoven and assigned it the role of sound recorder; the buildings create a score. Placed on a revolving wooden cylinder the buildings set little hammers in motion that play the keys of a piano. And turning and turning, the city makes its voice heard: from loud to soft, long to short, high-pitched to low, traslating the urban developers’s three-dimensional reality into an aural experience. Stadsmuziek (City Music) makes you tune in to the ensemble-playing that is environemental planning.

Via a terrific blog dedicated to data visualization (and, here, sonification) and their relation to design, Infosthetics. (Say that three times fast. Infosthetics sells aesthetic sensibilities by the seashore… Yesth, indeed. I’m glad this is a blog and not radio. It’s eight minutes past the hour. This is CDM.)

The player piano has come, oddly, full circle.

Peer into the editing pane of a music sequencer, and what you see is an abstracted virtual player piano. The editing views are even dubbed “Piano Roll” views (sometimes officially so). The MIDI protocol by which most music devices and apps communicate is itself a kind of port of the piano roll’s paradigm to a digital implementation, with fixed pitch and durations spanning time in an endless loop.

So it’s fitting that Dan Deacon is busy in the studio, feeding a player piano with MIDI and computers in order to make it play layers of unplayable music. The technical work is significant, to be sure, but on another level it’s natural that the computer would speak to the player piano in such terms. These are distant technologies bridged by a common understanding of how to simplify and reproduce music. The process speaks to some of the limitations in the way in which computers typically relate to musical instruments. We may be at the end of the road for this century-old way of thinking about mechanized music. But, extended to the near-breaking point, the maximalist texture that results is all the more beautiful.

Thanks to Jason Bergman for the tip.

Pitchfork TV: Dan Deacon in the Studio [Player Piano]