Unlocking Sounds of the Past

You may recall reading about some early audio recordings of Alexander Graham Bell being recently recovered. Thanks to Library of Congress technology, these recordings and others can now have a new life. The following is a story by Mark Hartsell, editor of the Library’s newsletter, The Gazette, on how the institution is working to accomplish this.

In the basement of the Madison Building, Peter Alyea sits at a desk and takes millions of photos of the insides of the grooves of old recordings.

Peter Alyea displays a digital image of the grooves of an old recordings captured by IRENE./Abby Brack Lewis

The images, reassembled on Alyea’s computer monitor, reveal in detail every curve cut into the floor of every groove – and provide the key to extracting sound from recordings unheard for decades and preserving those sounds for future generations.

The pictures are the product of IRENE/3-D, a system that uses digital imaging to retrieve sound from historical recordings made on discs and cylinders that might otherwise be unplayable.

IRENE allows damaged recordings – a disc broken into separate pieces, for example – to be reassembled digitally and for the sonic debris of pops, skips and distortions to be cleared away just as a technician might Photoshop a blemish from a photograph.

Over the past year, Alyea, a digital-conversion specialist at the Library of Congress, served on a multi-institution team of scientists and curators that used IRENE at the Library to extract sound from some of the oldest recordings ever made – a group of experimental discs produced by Alexander Graham Bell whose contents hadn’t been heard since they were made about 125 years ago.

The work is important to preserving the nation’s sonic cultural heritage.

A study released by the Library in 2010 found that major areas of America’s recorded sound heritage already have deteriorated or remain inaccessible to the public.

Most audio formats deteriorate over time: Discs warps, and both discs and cylinders break or degrade.

That’s where IRENE steps in.

“We know they’re going to fall apart, and there are things you can do to increase longevity,” Alyea said. “But we know they will go away, and we want to get the information off of them.”

IRENE is Born

The idea for IRENE was conceived when Carl Haber and Vitaliy Fadeyev, scientists at the Lawrence Berkeley National Laboratory in California, heard Grateful Dead drummer Mickey Hart discuss on a radio program the loss of cultural heritage through the chemical breakdown of recordings.

Why, they wondered, couldn’t we just make high-resolution images of a disc and use the pictures to extract the data embedded in the grooves? Why couldn’t software then convert that data into sound?

Haber and Fadayev experimented with the idea over 2003 and ’04. Eventually, a workable system was born and named IRENE – after the first record from which they managed to extract sound, “Goodnight, Irene” by the Weavers.

(The name then was reverse-engineered into an acronym: “Image, Reconstruct, Erase Noise, Etc.”)

IRENE uses high-resolution cameras to capture images of the grooves of recordings. Software then converts those images into sound./Abby Brack Lewis

With IRENE, a light shined onto a disc or cylinder reflects into a high-resolution camera mounted overhead. The camera makes images of the lateral motion of the groove – more than 2 million images for a typical 78 rpm recording. Software analyzes the motion of the grooves shown in those images, then converts the data into sound.

The system also helps preserve fragile historic material: Because a camera reads the grooves of a recording, the delicate discs are spared the wear inflicted by repeated physical contact of a stylus.

Alyea read a report about Haber and Fadeyev’s work and had his own eureka moment: He realized the approach could be used to extract sound from broken discs and that IRENE could be automated to produce digital copies of recordings in large numbers.

“All these things suddenly flew into my head when I read what they were saying,” Alyea said.

Alyea and Haber discussed the possibility of a collaboration between Berkeley and the Library, and the first device – fabricated at the lab – was installed at the Library in 2006. The original 2-D system later was installed at the Library’s Culpeper facility and a second 2-D/3-D version – better at reading the grooves of cylinder recordings – was installed at the Madison Building in 2009.

‘Ba-ro-me-ter, ba-ro-me-ter’

Over the past year, Alyea, Haber, Berkeley scientist Earl Cornell and Shari Stout and Carlene Stephens, curators from the Smithsonian Institution, worked at the Library to retrieve sound from a group of experimental recordings made by Bell at a Washington, D.C., laboratory about 125 years ago.

In the 1880s, recorded sound was a hot field of invention, and competition was intense among great innovators such as Bell, Thomas Edison and Emile Berliner.

Bell and his colleagues at the Volta Laboratory in the District tried many techniques and media to record sound: tin and wax cylinders; discs of metal, glass, plaster and paper; a tape of wax-coated paper strips wrapped around two reels – kind of a primitive tape recorder.

“Bell was working in this highly competitive, highly lucrative field of high-technology communications, much as people are expanding those boundaries today,” said Stephens, a curator at the Smithsonian’s National Museum of American History.

To protect himself against potential claims against his patents, Bell later sealed those experimental recordings – along with his lab notes and evidence of date, such as newspapers – in tin boxes and deposited them for safekeeping at the Smithsonian. He much later gave about 200 more of the recordings to the Smithsonian.

For decades, the recordings sat mostly untouched – and entirely unheard.

“Because the recordings were considered unplayable, they were placed in a remote part of the storage room and nobody was using them for research, for sound, for anything,” Stephens said.

In 2008, Stephens read about the Library’s work with sound recovery and contacted Library officials to discuss a joint project to extract sound from the Bell recordings.

The project team chose six recordings – deemed most likely to yield sound – to submit to IRENE.

“These are one-offs,” Stephens said. “They are recordings before a standard style of medium and recording technique was developed. Each one needs special attention. Each one is fragile.”

The team eventually was able to extract sound from each of the six recordings.

A recording made on a brass disc covered with wax yielded a recitation from “Hamlet.” A glass photodisc features the word “ba-ro-me-ter” enunciated over and over. Another glass disc yielded trilling, the date of the recording (March 11, 1885) and “Mary had a Little Lamb.”

This large glass photodisc contained the word “barometer” repeated over and over./Abby Brack Lewis

“It made the hair stand up on the back of my neck,” Stephens said. “It was so thrilling. It was so eerie. It was so much a glimpse into a time period we have never heard from before, ever.”

Saving 78s

With IRENE, the Library wanted not just a tool that could “play” previously unplayable recordings, but also a fairly automated system that would allow technicians to conduct a high-volume transfer of 78-rpm recordings to digital files without time-intensive work by operators.

“The model of being able to make fast copies non-invasively with staff not specifically trained to know all the ins and outs of audio engineering was very attractive,” Alyea said.

The goal was to produce in large numbers digital audio files that could be used for access and, in the best case, preservation – and eliminate the need to ever play a disc again.

That’s been a challenge: Recordings come in many different formats, shapes and sizes, and the set-up and software frequently would need tweaking to accommodate the multitude of types.

But technicians have succeeded in producing sound files good enough to provide the public access to the material without an engineer spending a lot of time on each recording: The system has been used to make digital copies of hundreds of rare recordings from the Library collections.

“It proved to be even more reliable than I thought it would be,” Alyea said. “I actually thought it would have more problems. They basically have no problems with the capture side.”

The work of extracting sound from the Smithsonian material presented the opposite idea: The set-up and software frequently were modified to suit individual discs or cylinders.

The reward, however, is to give listeners the opportunity to hear recordings that otherwise might never have been heard again.

“The prospect of being able to save these recordings, save the sound heritage, save the sound culture of civilizations all around the world is pretty exciting,” Stephens said.

More information about the Bell recordings – and clips of sound extracted from them – can be found here.  

More information about IRENE is available here.

2 Comments

  1. A Nikhwai
    March 12, 2012 at 1:17 pm

    It’s pretty amazing to read this article. It’s so exciting without even hearing the actual sound of the recordings. Awesome.

  2. Corey Bailey
    January 16, 2013 at 2:04 am

    I’ve been following this technology starting with the experiments at the Lawrence Berkeley National Laboratory. Hopefully, some day it will become available to the public sector because there is an immense amount of recordings in private collections that is not available otherwise. Meanwhile, perhaps the LOC could offer this technology to those who still have homemade records from the 1930’s and 40’s and those made in our USO’s during WWII and sent overseas to GI’s. There’s a lot of history there and as one who has transferred a few of these discs for their owners, I can tell you that those one-off records could certainly benefit from IRENE/3D.

Add a Comment

This blog is governed by the general rules of respectful civil discourse. You are fully responsible for everything that you post. The content of all comments is released into the public domain unless clearly stated otherwise. The Library of Congress does not control the content posted. Nevertheless, the Library of Congress may monitor any user-generated content as it chooses and reserves the right to remove content for any reason whatever, without consent. Gratuitous links to sites are viewed as spam and may result in removed comments. We further reserve the right, in our sole discretion, to remove a user's privilege to post content on the Library site. Read our Comment and Posting Policy.

Required fields are indicated with an * asterisk.