All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links.
If there is one thing computers do well, it’s math. All of music’s raw components — key, mode, melody, harmony and rhythm — can be expressed mathematically. As a result, computers can help people make music, even if they don’t know their elbow from an F clef.
The following apps for computer, web browser and smartphone put the power of artificially intelligent music creation in your hands or let you hear music that was created or manipulated by machines. Without further ado:
uJam
One of the most impressive demonstrations I’ve seen this year, uJam is the brainchild of longtime audio-software developers Peter Gorges and Axel Hensen and their celebrity partners Hans Zimmer (film composer for Dark Knight, Gladiator, Lion King) and Pharrell Williams (producer for Madonna, Shakira, Gwen Stefani). Their web app, yet to be released, crafts entire songs with precise accompaniment out of whatever the user whistles, hums or sings (or tries to sing) with an auto-tune feature to smooth out the rough spots.
You can use this Flash-based software to record cover versions of popular songs, but the real magic lies in creating something from scratch, either with your voice or a musical instrument, in a multitude of styles. Even if you only have a single instrument on hand, or just your own whistling lips, uJam can turn your output into a number of other instruments, as Gorges demonstrates below by turning a simple recorder melody into a full-on guitar jam.
Following the freemium model, uJam will be free to use on a basic level, with add-ons available for purchase.
Emily Howell
You can’t use the artificially intelligent Emily Howell software yourself. She belongs to her creator, University of Santa Cruz professor David Cope, who runs her on an ancient Power Mac 7500. However, you can catch her in concert from time to time at the university, when human players bring her compositions to life, in the iTunes music store, which sells her first album, From Darkness, Light, or in this embedded video.
Howell, a descendant of Cope’s earlier Emmy software (shorthand for his term “Experiments in Musical Intelligence”), works by creating connections between various musical statements and tonal relationships in order to churn out musical responses to phrases Cope feeds in — most of them coming from Emmy’s earlier scores.
The result is astounding to even the casual listener — rife with emotional complexity and deep textures that belie the music’s artificial origin.
Microsoft SongSmith
Microsoft, that dorkiest of companies, pioneered the concept of accompanying regular humans’ vocal warblings with artificially intelligent algorithms to create something that might, in some stilted universe, be called music.
Its SongSmith software (check out RobertSongSmith, or Microsoft’s notoriously atrocious commercial below to see just how bad it can get), tends to produce songs that sound like they were made with a Casio keyboard’s auto-accompaniment setting.
That said, it did manage to turn Metallica’s “Enter the Sandman” into an amusing disco number. SongSmith may have been first, but it’s very … well, Microsoft. Luckily, a new crop of artificially intelligent music makers has emerged since SongSmith launched early last year.
The Swinger
The Swinger (source code only) adds a playful rhythm to any song, because as any 1930s music enthusiast can tell you, “it don’t mean a thing if it ain’t got that swing.”
The Swinger is the work of The Echo Nest co-founder Tristan Jehan from the Hyperinstruments Group at MIT’s Media Laboratory. It’s designed not only for fun, but to demonstrate the power of The Echo Nest’s Remix API, which was also responsible for such gems as MoreCowbell.dj, a tool for adding cowbell and Christopher Walken samples to any song (as inspired by the legendary Saturday Night Live sketch), and DonkDJ, which transforms any MP3 into the Donk genre.
Sweet Child O’ Mine (Swing Version) by plamere
LaDiDa
Khu.sh’s “reverse karaoke” app LaDiDa has been turning heads with its earlier, mobile version of the uJam idea — an iPhone app that makes music out of whatever you sing in to it.
I shot a video (below) of the co-creators demonstrating the app at SXSW and explaining how it works — by analyzing what you sing and creating accompaniment around it. And because making music is no fun unless other people hear it, LaDiDa includes plenty of social features for posting the song to your Facebook wall, tweeting it and so on.
Robot Marimba
We covered Shimon, a robot marimba player created by the Robotic Musicianship Group at Georgia Tech’s Center for Music and Technology way back in 2008, but it bears mention again, having recently having made an appearance on The Colbert Report‘s ThreatDown list (video below).
According to Colbert, Shimon combines two great threats to America — jazz and robots — but it’s no joke. Somewhat along the lines of Emily Howell, Shimon applies its own melodic, harmonic and rhythmic sensibilities to input from a human collaborator in order to create new, artificially intelligent music.
The results are so impressive that we declared this and Georgia Tech’s other National Science Foundation-funded music robot, which plays the tabla drum, to have passed the musical Turing test (by having the potential to fool human listeners into thinking that its improvised responses to human-created music were those of another human).
Honorable Mention: Auto-Tune
Although it’s clearly responsible for some of the worst songs ever recorded, Auto-Tune, claimed by its developer Antares Tech to be “the largest-selling audio plug-in of all time,” has to be the world’s most widely implemented collision of the human voice and the computers trying to make sense of it. LaDiDa and uJam each rely on a version of the technology for lining up the human voice with their automatic accompaniments, and “Auto-Tune the News” employs it to hilarious effect, proving that Auto-Tune can be used for good as well as evil.
