Bringing Forgotten Music Back to Life with Optical Music Recognition

Just as printed letters can be scanned and converted into digital text, emerging technology can transform sheet music into sound.

A new effort is underway to use optical music recognition (OMR) to bring to life music in the public domain for all to listen and access through the Internet Archive.

Nearly 200 public domain songs have been uploaded to the Internet Jukebox —from John Philip Sousa marches to children’s folk songs and piano concertos by Frederic Chopin—and the collection is growing.

Marty Lucas and Carl Malamud spearheaded the project through the nonprofit organization Public Resource, with funding from Arcadia, a UK charitable fund. The two were college roommates at Indiana University in the 1970s and have been long-time work collaborators who shared a love of music.

This undertaking began about three years ago when Lucas, an attorney with an expertise in sound engineering and a fellow with Public Resource, discovered old sheet music in his farmhouse in Indiana. He suggested it could be scanned and preserved. Malamud introduced him to the OMR engine developed by Soundslice, a music instruction site, and the project took off.

Lucas converted printed sheet music to a digital format typically using MIDI software for playback and Music XML, a mark-up language for page layout. Then, he exported it into a digital audio workstation that allowed him to assign performances to particular instruments and mix. He started with sheet music he owned, digitizing and converting others in the public domain he found through libraries and online.

While a musical genius may be able to look at a page of sheet music and hear it in his or her head, Lucas said, the technology makes it easier for someone to actually listen to how it was originally intended to be played.

“It’s really exciting because it opens up the door to lots of different kinds of music—providing a new kind of access, very inexpensively,” Lucas said.

Sample some of the greatest hits in the collection:

SILENT FILM THEATER ORCHESTRA –Marceline (George J. Trinkaus) –Heavy Tension (Leo Kempinski) –The Vampire (Sol P. Levy) –Uneasiness (Gaston Borch)

VINTAGE POP –It’s A Long, Long Way to Tipperary (ft. John Kennedy Jr.) –Dance of the Daisies (LaFarge) –Alice Blue Gown (ft. Corinne)

SOUSA –The Gladiator (JP Sousa) –The Pathfinder of Panama (JP Sousa)

FOLK –The Harp that Once Thro’ Tara’s Halls (Thomas Moore) –How Should I Your True Love Know? (Anonymous)

CLASSICALPassepied (From the Fifth English Suite) (JS Bach)Lieder ohne Worte Opus 30, No. 5 in D Major (Mendelssohn)

Many pieces in the collection were played by pit orchestras during the early days of silent films (called photoplays at the time) from about 1910 to 1923.

“During this heyday period, people went to the movies to hear the orchestra because they didn’t have to dress up fancy. They could pay 50 cents and hear this beautiful music,” Lucas said, noting the music blossomed then withered away once theaters shifted to using organs. “There aren’t many recordings of these pieces and they’re really worthy of hearing. These are by some of the best composers of the day. It was neat to bring back pieces of quality music that hadn’t been heard in a century.”

The Internet Jukebox can serve as an educational tool for individuals teaching themselves or band directors working with students section by section about the sound of a piece. The collection could also be useful to people with videos looking to put a score with their content or musicians looking to recreate and remix.

“We’re hoping the creatives of the world who write songs and do music videos will figure out how it works,” Malamud said. “It has amazing potential. The key is to make this available and then see what kind of magic happens.”

To inspire the public to add to the collection as citizen archivists, Lucas and Malamud recently released a series of instructional videos explaining how the OMR process works.