Reviving and Restoring Lost Sounds

In 02008 Kevin Kelly called for movage (as opposed to storage) as the only way to archive digital information:

“Proper movage means transferring the material to current platforms on a regular basis— that is, before the old platform completely dies, and it becomes hard to do. This movic rythym of refreshing content should be as smooth as a respiratory cycle — in, out, in, out. Copy, move, copy, move.”

Five years later, Berkeley physicist Carl Haber received the MacArthur “Genius Grant” for doing just this–moving two- or three-dimensional audio recordings on obsolete platforms and/or decaying storage media and digitally restoring them. These long-lost analog sounds can essentially be played with a virtual needle.

Haber already had the technology in place from his research on imaging radiation. He had cameras precise enough to image and measure the patterns of particles and debris that emerged from subatomic particle collisions. He and his colleagues at the Lawrence Berkeley National Laboratory applied this noninvasive image processing to develop IRENE (Image, Reconstruct, Erase Noise, Etc.). They derived the acronym from the first recording used to demonstrate the concept of IRENE–The Weavers performing “Goodnight Irene.”

Just like the detailed technique required to measure radiation, pictures taken at great magnification are needed to map the surface of an audio recording.  One pixel is about one micron on the disc or cylinder surface, meaning in order to acquire a sufficient digital map, the camera scans the object slowly enough to synthesize a gigapixel image:

A disc or cylinder is placed in a precision optical metrology system, where a camera following the path of the grooves on the object takes thousands of images that are then cleaned to compensate for physical damage; the resulting data are mathematically interpolated to determine how a stylus would course through the undulations, and the stylus motion is converted into a standard digital sound file.

So IRENE uses image processing to take a picture and mathematically break down the information in that image to calculate the motion of the groove and determine what sound would actually be played. It’s all done with an algorithm on a computer, never having to touch the recording in the process.

Haber has collaborated with archivists and researchers around the world to test IRENE on a variety of audio recordings. The Smithsonian has about 200 experimental recordings from Volta Laboratory Associates, a collection of some of the earliest audio recordings ever made. It is “a reflection on the intense competition between (Alexander Graham) Bell, Thomas Edison and Emile Berliner for patents following the invention of the phonograph by Edison in 1877.”

Little-Lamb-glass-plate-no-3-analysis
A glass disc recording from Bell’s Volta Lab containing the audio of a male voice repeating “Mary had a little lamb.” Photo: The Smithsonian

Now, in collaboration with the Smithsonian and the Library of Congress, IRENE has started to recover these fragile recordings made out of rubber, beeswax, glass, tin foil and brass. The cryptic recordings on such delicate and damaged storage mediums can presently be heard, over a century later:

Early experimental recordings are not the only recovered lost voices. Anthropologists, linguists and ethnographers were among some of the first to use recording as a research tool and to document cultural heritage. IRENE has restored some of these field recordings, including wax cylinders from the Alfred L. Kroeber collection at the Phoebe A. Hearst Museum of Anthropology in Berkeley. There are over 3,000 cylinder recordings in the collection that document California Native American culture from 01900-01919. Around 300 of these cylinders are 2-3 minute long recordings of Ishi, the only surviving member of the Yahi at the time. 53 of the 300 cylinders are Ishi telling the story of the “Wood Duck” in 01912.

Earlier this month Smithsonian’s National Museum of Natural History received a $1 million grant from the Arcadia Fund to digitize its endangered-language materials, including the estimated 3,000 hours of sound recordings. The recordings will then be electronically available to the public through the Smithsonian’s catalog system:

Digitization of these materials within the NAA (National Anthropological Archives) will give both scholars and local communities new access to documentation of endangered languages and cultural knowledge about threatened environments around the world, ranging from southern California to small Micronesian atolls.

In September 02010, The Library of Congress released the first comprehensive study on a national level examining the preservation of sound recordings in the United States. It found that many historical recordings have already deteriorated or are inaccessible to the public due to their experimental and fragile nature.

“Those audio cassettes are just time bombs,” the study’s co-author Sam Brylawski said. “They’re just not going to be playable.”

But maybe this is not the case after all. Perhaps Haber has taken upon an archaeological endeavor that postpones the detonation of media “time bombs.” It comes back to Kevin Kelly’s idea of movage. Haber has taken big technological steps to digitally play these long-lost analog voices, now the key is to keep on moving.

Share on Facebook Share on Twitter

More from Archives

What is the long now?

The Long Now Foundation is a nonprofit established in 01996 to foster long-term thinking. Our work encourages imagination at the timescale of civilization — the next and last 10,000 years — a timespan we call the long now.

Learn more

Join our newsletter for the latest in long-term thinking

Long Now's website is changing...