The McGill Image-to-Audio Conversion Project (MItAC)

The McGill Image to Audio Conversion (MItAC) research project focuses on the access to and the long-term preservation of the phonograph sound recordings. There are a large number of damaged phonograph records that are unable to be played or digitized by using the traditional turntable systems. An alternative approach is to optically scan the phonograph disc to create the 2D or 3D digital image of a damaged record and subsequently convert the scanned image to audio for access. Because the standard digitization and preservation workflow for phonograph records are still under exploration, this approach, which is often called optical audio reconstruction (OAR), has the advantage of minimal physical contact with the records during the experiments with various digitization and sound restoration techniques.

There have been other relevant research efforts in OAR conducted by Lawrence Berkeley National Laboratory in California and University of Fribourg in Switzerland. These efforts are only towards the mono phonograph records, namely the 78rpm or the wax cylinder records. See Previous Work in Optical Audio Reconstruction.

Our research, in contrast, aims mainly at the restoration of the 33rpm stereo recods. The measurement equipment that we use is a microscope that adopts white-light interferometry, the Wyko NT8000 Series Profiler, which is controlled by the Wyko Vision Software, both purchased from the Veeco Instruments. See System Hardware and Software.

The Documentation consists of the analyses of specific issues and the descriptions of algorithms developed with the links to the source codes (Algorithms and Analyses), the geometric information, metadata, and other information of the phonograph records that are relevant to our experiments (Disc Information), the data format specifications used in all our experiments and analyses and the utilities developed by us to facilitate the measurements and analyses (Data Formats and Utilities), the documentation of the optical profiler and its software usage (System Hardware and Software), along with the screenshots, audio, and video resources of our experimental results and demos (Multimedia). Various specialized terms, mainly from the profiler specifications and our inventions, are used in all the documentations. They are clarified in Glossary.

The Experiments keep the records and the data of the important experiments that we made, including the measurements that lead to major results (Measurements), the exploratory tests regarding the functionality and mechanics of the hardware and software of our profiler (System Functionality and Performance), and all the temporary data obtained from experiments (Temporary Data).

The Annotated Bibliography lists all relevant literature on Optical Audio Restoration, Phonograph, and Interferometry with our brief highlights.

The Links point to the resources outside this Wiki, including the work blogs of our Team, which serve as daily notifications of the progresses and updates on the Wiki, and the bug reports of the Wyko Vision software.

Mainly supported by Daniel Langlois Foundation Grant, this project proceeds in parallel with the digitization of McGill music library's 78rpm Jazz recording collection (funded by a three-year FQRSC research grant) and digitization of a unique collection of Handel LP recordings, the David Edelberg Handel LP Collection (funded by McGill's Richard M. Tomlinson Digital Library Innovation Awards).

For more information of us, see Team.