Tuesday, December 8, 2015

Sound Imaging For Preservation and Access to Historical Recorded Sound


INTRODUCTION
As our cultural heritage continues to be made up of more and more audiovisual forms of expression, our archives are becoming filled with historic audio recordings that we, as archivist, must preserve and make accessible.  Recordings of our musical artists, poets, writers, news, speeches & spoken words of our politicians and great thinkers, radio broadcasts and field recordings of many cultures, private and public dictation, intelligence recordings and all commercial record releases make up just some of the major categories of our historical recordings.  

The Recording Preservation Act of 2000  made sound preservation a matter of national policy to “maintain and preserve sound recordings and collections of sound recordings that are culturally, historically, or aesthetically significant… " (Public Law 106-474; H.R.4846).  A new advancement in the preservation of audio has emerged called Sound Imaging, which tackles the preservation of audio carries that are made from grooved audio media such as cylinders and discs and the first 70 years of our audio heritage. This advancement employs the use of photosensitive sensors in optical probe cameras to acquire data images of audio carriers, which can be translated into sound.

A BRIEF HISTORY OF SOUND IMAGING
The first instance of sound imaging is credited to Édouard-Léon Scott de Martinville or Leon Scott, a French printer and bookseller from Paris who patented the first known sound recording device in 1857. He called his machine the Phonautograph and the result created a visual image of sound on paper.   Sound traveled through the horn of a cylinder that vibrated a diaphragm at one end of the horn. A stiff brush attached to the diaphragm would move across a rotating cylinder covered with lampblack or blackened paper from the sound vibration.  This created an image of the sound. He also used flat discs to trace the lateral motions of the stiff brush. However, there was no way to playback these “recordings” be they disc or papers.

Thomas Edison was the first inventor to create the “Phonograph” machine that actually played back sound in 1877. He used a tinfoil rather than paper to emboss the audio signal and a stylus to playback or retrace the waveform. Charles Summer Tainter and Alexander Graham Bell developed the “Graphophone” from 1881-1886, which used a wax like cylinder as its audio carrier. Emile Berliner invented the “Gramophone”, in1887 using non-wax “metal plates” or discs as the carrier which was photoengraved with lateral grooves. The race to invent better mechanical recording devices and their audio carriers had begun.

The “metal plates” became playable discs or records that we can listen to over and over again.  The audio recording and playback industry is now over 130 years old.


   Scott’s Phonautograph 
    











             Audio tracing on embossed on paper              
Edison’s Phonograph
The audio cylinder continued to evolve utilizing different sizes and physical properties. Eventually 78rpm records became the standard for commercial audio carriers followed by the “micro grooved“ LP records we are still listening today- despite the advancements of optical media such as audio CDs or digital file media such as audio wav and mp3 files.

The problem for archivist is how to play back these grooved audio recordings without damaging them and how to play back recordings that are damaged and sometimes broken. Thousands of recordings were unavailable because of these kinds of conditions, prompting Congress to stress the need to develop new methods of preserving audio.

Sound imaging is a new method to hear the “sound” embedded in a grooved sound carrier.  No stylus or needle is used in this method and thus no damage or “playback wear” is done to the existing audio carrier by the stylus. The grooved surface of the sound carrier is imaged and then translated into audio through the use of specialized camera and that can take very small images or photos of the grooves. A high intensity laser beam is shown on the surface of the audio carrier. Those images are then translated into a digital audio wav that can be played back for humans to listen to. In the case where the audio carriers are damaged or in pieces, sound imaging can image the parts and knit them back together in software to again be heard.

Although still in its infancy, sound imaging had its’ early roots in the 1960’s when reflected light patterns began to be used to characterize test records and researchers studied grooves with infrared photo-metry in the early development of the compact disc or audio CD. Laser light improvements developed throughout the 1970s with the first laser made CD player coming out in 1980. In the 1990’s a laser turntable was developed in Japan and throughout the 2000’s laser reflection was tested to replace the traditional styli with light.

VisualAudio is the first optical audio signal retrieval system based on analog high-resolution photography and subsequent fill scanning and was developed by Dr. Stefan Cavaglieri at the College of Engineering and Architecture of Fribourg, Switzerland.  Beginning in 2001, their aim has been to archive discs that are deteriorating by storing their scanned sound images on photographic film that support a long life span.

Dr. Carl Haber of University of California Berkley Lawrence Livermore Labs heard a piece on public broadcasting about the loss of historic recordings and thought his work with precision optical metrology systems for the ATLAS experiment in Cern, (which observed the particle called the Higgs Boson) could become the basis for inventing a system to scan grooved audio carriers.   Optical metrology uses “electronic imaging, image processing and precision motion control to derive quantitative information about the size and shape of physical objects”. [i] The basic research for this work was carried out in 2002-2004.  Dr. Haber called his imaging system IRENE  (Image Record Erase Noise, etc) and with Dr. Earl Cornell has developed the IRENE imaging process.

In 2005 the National Endowment for the Humanities funded further development of disc imaging for use at the Library of Congress and the first disc imaging system was installed at the LOC in 2006. The Institute of Museum and Library Services funded a 3D imaging system for audio cylinders that was installed in the Library of Congress in 2009.  Follow up grant funding by IMLS from 2009-2012 helped create more systems for imaging more collections, notably one in India for disc imaging, and carried out a series of special projects on small collections that let us hear audio from the earliest days of recorded sound, creating significant access to the research community in particular. Many other institutions supported this research including the National Archives, The Department of Energy, The Mellon Foundation, the Smithsonian Institution, etc. See index listing for some of these special historical audio projects. The first imaging system that made this system available to the public began serving the preservation community at Northeast Document Conservation Center in 2014.

A recent Sound Imaging Conference in July of 2015 at the Library of Congress highlighted the advancements in imaging currently underway in Britain headed up by Dr. John McBride and by Sweden’s Dr. Ottar Johnson, among others.
HOW IT WORKS
Sound Imaging is a general term used for profiling the surface of grooved mechanical sound carriers through the use of high-resolution digital microphotography to capture a representation of the grooved disc surface. Called optical surface profiling, microphotography captures or “maps” the sound carrier surface, creating 2 dimensional or 3 dimensional data images. 
HOW IT WORKS
The IRENE has two methods of doing this for two different grooved audio carriers. Two-dimension microphotography mapping is typically used for laterally cut sound carriers such as disc or records. The imaging is based upon high-resolution digital microphotography and captures a two-dimensional (2D) representation of the disc surface. The aim is to address the high-speed capture of the revolving disc. A light source is needed on the surface to illuminate the grooves for the camera and a laser sensor measures the distances of the audio carrier’s grooved surface.
The camera takes image of record as it turns, imaging the grooves
              

Three-dimensional mapping is used for vertically cut sound carriers such as cylinders.  A confocal micro photographic camera, as used in microscopy, takes 180 points of the depths on the audio carrier’s grooves. A laser beam measures the distances of the audio carrier’s grooved surface in order for the camera to focus correctly on the microscopic surface.
The camera takes multiple points of the cylinder’s surface, resulting in black and white data file of cylinder’s surface.
                       
Once high-resolution mapping is made of these 2 kinds of grooved sound carriers, computer-processing methods are applied. The edges of each groove are found by focusing on areas of high-contrast between light and dark in the image. The flat bottoms of the grooves and the spaces between tracks appeared white on the data images, while the sloped sides of the grooves, scratches, and dirt look black. The data image is analyzed to create a digital audio wav form for listening.  The preservation master is the data file of the image; the access master is the audio file.
           
View of Grooved disc or ‘record’
A numerical or mathematical algorithm, executed on this mapped image of the sound carrier, reconstructs the recorded sound with no contact of the actual sound carrier. This maintains the original sound carrier’s physical properties without anything actually touching the recording medium, (and possibly degrading it) creating a new method of preservation for these sound carriers. Additionally by mapping the damaged or broken parts of these sound carriers it is possible to actually “hear” what is on them again. Librarians, archivist, museum curators and cultural historians have been waiting for an opportunity to be able to recover these damaged and “unplayable “ recordings. With sound imaging, we now have the possibility of doing just that.

Some of the issues surrounding the advancement of sound imaging are speed in transfer time for this process, lack of trained audio personnel to utilize this system, and lack of further development money to make the system more standardized for audio engineers or “user friendly”. Multiple-cylinder scanning, now 3 cylinders at a time, is currently under way and NEDCC is training multiple users on the system in 2016. Although in its infancy, sound imaging is an audio preservation development that will be advanced and utilized in the future in ways that may be difficult to imagine.

Works being done at MIT with microscopic cameras, even regular cell phone cameras, are imaging and reproducing sound recorded in objects, giving way to the idea of embedded audio in the all grooved objects in our collections. It is hoped that soon we will be able to start retrieving audio embedded in our collection’s objects- giving way an even greater audio collection and cultural heritage. For instance we could listen to a ceramic pot that had Latin being spoken when it was made and hear this “dead” language, as it was spoken thousands of years ago. I urge everyone to watch the 4 minute video explaining some of MIT’s sound imaging current work. 

CONCLUSION
We have steadily gone from the evolution of sound recording and photography as separate entities to a blend of both technologies that document our cultures. It is up to archivist to preserve and to create access  these artifacts from the evolving sound imaging pathways that allow us to do so.



Bibliography
Cornell, E.W. ,.Fadeyev,V., Golden,M. .Haber,,C.,Nordmeyer, R.(2009). “3D Optical Scanning of Mechanical Sound Carriers Technical Description RevisedRetrieved from http://irene.lbl.gov/IRENE_2D_3D_Examples.html

Davis, A. (2014). The Visual Microphone: Passive Recovery of Sound from Video

Watch: https://www.youtube.com/watch?v=FKXOucXB4a8&feature=youtu.be

Haber,C. (2014). “New Technologies to Preserve and Access Historical Recorded Sound
June 2013”.  Retrieved from: Sound-Project-0511s.pdf

Hardesty, L. (2014) “Extracting Audio from Visual Information Algorithm recovers speech from the vibrations of a potato-chip bag filmed through soundproof glass.”. Retrieved from:http://news.mit.edu/2014/algorithm-recovers-speech-from-vibrations-0804

Holton, Conard  C. (10-01-2007). “Rescuing recorded sound from silence”. Retrieved from http://www.vision-systems.com/articles/2007/10/rescuing-recorded-sound-from-silence.html

IRENE Sound Reproduction R&D Home Page Irene.lbl.gov  
General information

Johnson, O.,  Sotozer, S.. Bapst, F., Ingold, R. (2007).  “Detection of the Groove Position in Phonographic Images”. Fribourg Univ., Fribourg Conference: Image Processing, 2007. ICIP 2007. IEEE International Conference on, Volume: 6 Retrieved from https://www.researchgate.net/publication/4289187_Detection_of_the_Groove_Position_in_Phonographic_ImagesDOI

Krotz, Dan (2004). “From Top Quarks to the Blues:Berkeley Lab physicists develop way to digitally restore and preserve audio recordings” Retrieved from
http://www2.lbl.gov/Science-Articles/Archive/Phys-quarks-to-blues.html

Schoenherr. S.E. (2004). “Leon Scott and the Phonautograph”. Retrieved from http://www.aes.org/aeshc/docs/recording.technology.history/scott.html
Reference:
To listen to recovered archival sound check out: “Examples from the Berkeley Lab Recorded Sound Restoration Project” Last Update 23-September-2013.  Retrieved from: http://bio16p.lbl.gov/2013-examples.html

Footnotes:


[i] Haber,C. (2014 )“New Technologies to Preserve and Access Historical Recorded Sound
June 2013”.  Retrieved from:

No comments:

Post a Comment