Imaging Voices: Optical Scanning Applied to Recorded Sound Preservation and Access Carl Haber Lawrence Berkeley National Lab. 15-Jan-2014 SI NMNH 1
Edison s Invention In 1877 Edison was experimenting with methods to record telegraph impulses on paper discs. This gave him the idea of recording the continuous impulses of the voice permanently in a material.. 15-Jan-2014 SI NMNH 2
1878 foil 50 µm 1 mm 75 µm 400 µm 150 µm ~1940 disc. 15-Jan-2014 SI NMNH 3
Formats Early recording is dominated by mechanical (grooved) formats. Grooves moved either side-to-side or vertically The earliest recordings utilized diverse materials and configurations, particularly in the experimental period.. 15-Jan-2014 SI NMNH 4
Contingency Fortunately, much evidence of early sound recording has been preserved in collections. The sound carriers, the machines, the notes Unfortunately many of early carriers are at risk and traditionally problematic or even unplayable Carriers are delicate, damaged, playback systems don t, or may never have existed More generally archives want to digitally preserve their extensive collections - ASAP. 27-Feb-2013 CITRIS 5
A Non-Invasive Restoration Could we optically digitize a recording without contact to the medium like robotic text scanning? Address concerns of the preservation, archival, and research communities: Preservation: Restore or stabilize delicate or damaged media Access: Mass digitization of diverse media, automation Assessment Obsolete formats and legacy playback systems: generality Precision optical measurements are widely practiced in the physics and engineering research lab.. 15-Jan-2014 SI NMNH 6
A Basic Optical Process High resolution optical probe creates a series of depth/intensity profiles of the surface 20-80 mins C Create audio waveform These are merged into a surface map Calculate the motion of a virtual needle, apply optional restoration Map is archived. 15-Jan-2014 SI NMNH 7
160 µm 2D Imaging for Lateral Grooves scratch dust surface groove bottom Coaxial illumination h Require 1 pixel = ~ 1 micron on the disc surface High resolution = narrow depth of field, 10 20 microns Requires active auto-focus control High speed cameras allow near real-time imaging Extract groove information from high contrast edge transitions IRENE. 15-Jan-2014 SI NMNH 8
3D Imaging: Confocal Microscope Time Vertical resolution ~50-100 nm 0.05 mm. 15-Jan-2014 SI NMNH 9
System Implementation. 15-Jan-2014 SI NMNH 10
Analysis Software Analysis parameters and options Overview of full data set Profile along red line Zoomed in view Depth image, black is deepest Surface damage The analysis package PRISM includes powerful tools and options for access to the data and image processing to remove defects and damage.. 15-Jan-2014 SI NMNH 11
Dust Removal Dust particles appear WHITE because they are above the surface. 15-Jan-2014 SI NMNH 12
Examples Broken shellac disc ~1930 Contemporary lacquer cut disc You will soon go to a ball or large gathering and meet a new friend. A sincere friend seeks to help you in matters of importance to you. Your troubles can be avoided by changing your attitude towards them. Cracked wax cylinder 1906, fortune teller Earliest surviving disc record 1881, AG Bell & CS Tainter Trr, 1 2 3 4 5 6 Trr, Trr. 15-Jan-2014 SI NMNH 13
Applications and Projects Beginning in 2003, this approach has been developed into working systems, in collaboration with the Library of Congress Tools to handle diverse materials and pilot studies May apply particularly to field collections and at risk items Tools to process large collections, workflow Faster and robust hardware and software Method and approach developed with/for Library of Congress System recently installed at Northeast Document Conservation Center Other installations are under discussion The history of a technology Have now rendered playable examples of all the historic milestones in the development of recorded sound (~pre-1895). 15-Jan-2014 SI NMNH 14
Field/Transcription Collections Media Approx Years Example Collections How many? Example Challenges Wax cylinder 1890-1940 s Native American Aluminum disc 1925-1935 Parry, Lomax, Harrington 50,000 breakage delicate mold 10,000 tiny groove oxidation Lacquer disc 1935-1965 Radio, Studio Many x 10K exudation flaking Dictation belt 1947-1970 s Presidential Many x 10K tiny groove cracked. 15-Jan-2014 SI NMNH 15
Field Cylinder Pilot Studies Worked with 2 collections Have scanned 60 items from UCB Hearst Museum representing a survey of Native Californian materials (~1900-1914) Have scanned ~20 items of Kwakiutl material recorded by Franz Boas on Vancouver Island in 1930 (Indiana ATM, UW Burke Museum) Can we create improved access to these materials? Measure and develop a project workflow Correlate/synchronize with motion pictures (Boas filming) Boas #711. 15-Jan-2014 SI NMNH 16
Overcut Cylinder 711 Stylus version (blue) 3D version (green) Since optical scanning is free from the real-time dynamic effects inherent in stylus playback certain types of distortion can be reduced. 15-Jan-2014 SI NMNH 17
Phoebe Hearst Museum Pilot Digitization Study 2011-2012: 60 cylinders (of 3000) transferred by Maryrose Barrios (UCB physics student) 20/week, developed measurement and analysis parameters Created database and posted results Presented at 2012 Breath of Life. 15-Jan-2014 SI NMNH 18
Native American Field Recordings at the UCB Hearst Museum ~3000 cylinders by Prof. Alfred Kroeber and associates beginning ~1900 Yurok Gambling Song Central Sierra Miwok Yahi Ishi Recordings Wood Duck Story Stylus playback Giant s Song. 15-Jan-2014 SI NMNH 19
Broken Wax Cylinder Temporarily constrain pieces on the mandrel with plastic straps and (re-useable) putty, shift strap, scan in segments. 15-Jan-2014 SI NMNH 20
Jack London Dictation Cylinder (J L State Park) December 2, 1915 Dear Max Ehrmann: Just a rush letter, ere I sail for Hawaii. I merely want to tell you that everything concerning California prisons in the Star Rover is true. Ed. Morell is a real man, and Ed. Morell is his real name. He had a fifty year sentence, and he spent five years of it in solitary, as I have described. Two years ago Jake Oppenheimer was executed in California for assault and battery. I can only repeat, that what I have described is true of California up to the year 1913. I do not know what has happened in California since that date. If you ever read a book of mine entitled The Road, in which I give some few of my experiences, you will notice that in the Erie County penitentiary at Buffalo, New York, I have slipped by without describing much of the worst that I found obtaining there. What I found there was unprintable, and almost unthinkable. I am still curious to know how my handling of the Christ situation in Jerusalem will strike you.. 15-Jan-2014 SI NMNH 21
Aluminum Discs Aluminum discs were in use from mid 1920 s-30 s Grooves were embossed in the soft metal, considered noisy and loud Grooves were very shallow as compared with shellac or lacquer They were meant to be played back with fiber stylus so as to not score or damage the surface Study done in collaboration with Milman Parry Collection of Oral Literature at Harvard University. 15-Jan-2014 SI NMNH 22
Aluminum Disc Study Typical groove depth* 4147 Stylus 3D Optical (* of a commercial shellac 78 rpm or lacquer disc). 15-Jan-2014 SI NMNH 23
Lacquer Transcription Discs Introduced in the mid-1930 s as a replacement for aluminum. Much improved audio quality but susceptible to delamination, flaking, and chemical breakdown 78 rpm acetate Theos Bernard, interview, 1930 s (UCB) 78 rpm nitrate on glass Label: Howard Hughes, Collier Award 1939 (Lakeland Mus.) 78 rpm acetate on metal 1940 s studio test Mutt Carey and the NY rs (LoC). 15-Jan-2014 SI NMNH 24
Broken Disc Strategy. 15-Jan-2014 SI NMNH 25
Interactive Restoration Tools. 15-Jan-2014 SI NMNH 26
Plastic Dictation Belts Dictation, telephone and radio monitoring 1940 s-1970 s US Presidential phone conversations Groove is embossed, lateral modulation, shallow Never meant as an archival medium, poor storage. 15-Jan-2014 SI NMNH 27
Production Scanning Tools Developed in collaboration with the Library of Congress / Culpeper Disc data entry Links automatically to database and index file Operator quality and comments entry Minimal run commands and settings Scan status, progress bar, and task queue Image quality. 21-Mar-2012 IMLS 28
IRE # Condition data RENE New images? Process Resample Warped? in/out? autopos N autofocus Image quality Scan Results database. 15-Jan-2014 SI NMNH 29
Database for Production Scanning IRE00412 The World Today 1942-05-94 Audio Initial focus quality Label shot Image analysis Focus control. 15-Jan-2014 SI NMNH 30
History of a Technology 1895: Dickson Cylinder film soundtrack 1888-1895: Emile Berliner Gramophone disc 1887: Edison s talking doll commercialization 1881-1885: Volta Laboratory developments 1877: Edison records and reproduces on foil 1860: Edouard Leon Scott de Martinville records sound on paper Phonoautograph. 15-Jan-2014 SI NMNH 31
The Volta Laboratory In 1880 Alexander Graham Bell established the Volta Laboratory at 1221 Connecticut Avenue, Washington, D.C., to conduct research on sound recording and other topics. He formed an association with chemist (and cousin) Chichester Bell and instrument builder Charles Sumner Tainter. The associates experimented with an astounding variety of materials and formats. They produced numerous patents before settling on the wax cylinder as a recording medium of choice. Most of the experimental materials and notes are now in the collections of the Smithsonian Institution.. 5-Dec-2013 ASA 32
Wax on binder board: AG Bell speaks 60 µm This record has been made by Alexander Graham Bell, in the presence of Dr. Chichester A. Bell, on the 15 th of April, Eighteen hundred and eighty five, at the Volta Laboratory, 1221 Connecticut Ave, Washington, DC, in witness whereof, hear my voice, Alexander Graham Bell. 5-Dec-2013 ASA 33
Graphophone (1881) Trill, trill, There are more things in heaven and earth, Horatio, than are dreamt of in our philosophy. Trill. I am a graphophone and my father was a phonograph. Recording of Melville Bell (AGB s father), September, 1881 Bell modified an Edison tin foil machine to use wax instead. For patent rights It was sealed in a metal box and deposited at the Smithsonian.. 5-Dec-2013 ASA 34
The St. Louis Tinfoil The earliest Edison recording yet restored. It was recorded in June of 1878 as part of a public demonstration of an early commercial machine. The event was staged by journalist Thomas Mason. The foil passed through a number of collectors and was finally given to the misci in Schenectady, NY. The foil had been folded with 7 distinct creases.. 5-Dec-2013 ASA 35
. 5-Dec-2013 ASA 36
click click click 3s Unidentified brass 24 s Mary had a little lamb 10 s Laughter. 9s More laughter Old Mother Hubbard, went to the cupboard 11s Full recording is 1 min, 10 seconds, consisting of 6 distinct sections Full filtered = remove extra frequencies. 5-Dec-2013 ASA 37
The Phonoautograph Edouard Leon Scott de Martinville 1853-60 Literally, sound self writer Scott was inspired by the demonstration of photographic capture of images on paper. His perceived application was stenography. Yet he had the vision to record 2 simultaneous tracks in order to provide a real time calibration of the hand cranked process. 1 st digitized (by FirstSounds) and restored in 2008. 5-Dec-2013 ASA 38
2008 1860 "Au Clair de la Lune" ["By the Light of the Moon"] sung; the pitch is measured by the tuning fork of 500 simple vibrations per second which writes directly and simultaneously in interlinear space of the song (French Patent Office) Léon Scott 9 April 1860. 5-Dec-2013 ASA 39
Collaboration and Support Lawrence Berkeley National Lab The Library of Congress The Smithsonian Institution Univ. of Appl. Sciences, Fribourg, Switzerland Northeast Document Conservation Center Earl Cornell, Peter Alyea, Carlene Stephens, Shari Stout, Ottar Johnsen. 15-Jan-2014 SI NMNH 40
Conclusions Optical technology and processing can aid archives, museums, and libraries in preserving and gaining digital access, noninvasively, to at risk materials, collections, and historic media. Approach is very general Methods are in use in support of a variety of collections. Access to this technology exists and is expanding. For more information: http://irene.lbl.gov/. 15-Jan-2014 SI NMNH 41