X

Week 1: Starting the Weaver project, loading image, etc

Mar 12, 2021

Hello everyone, my name is Leo. My senior project, titled “Development and implementation of new computer processing methods to extract sound from images of historic grooved audio media”, is conducted in the Library of Congress (online) under my project mentor, Mr. Peter Alyea from the PRTD division of the Library of Congress, and my faculty advisor Dr. Ryan Grove.

Project Summary

I will be working on the software development concerning the image and sound
processing on the Weaver software, which is depended by the Library of
Congress’s Irene System, a system that high-resolution images of historic grooved audio media and subsequently processes the images to extract the sound.

Starting with the project

Majorly written in the programming language C#, the Weaver project could be
developed using Visual Studio, an integrated development (IDE)
by Microsoft. After opening the source code of Weaver using Visual Studio, we can run the project and use the weaver to load images.

Loading the first image

We can load the image by using the coding function ‘LoadTiff’ from the code. In loading the picture, we need to make sure that the extension of the picture is
Tiff. The difference between Tiff file type and other file types, such
as JPEG, PNG, and GIF, is that the file doesn’t compress, and thus is better for
editing. Loading a tiff image by running the ‘LoadTiff’ plugin in the system
results in the following image.

pic-result

Binning

Because of the big size of the scanned audio media, when only generating the visual and tracking it, we would only take a smaller size of it for display and analysis. This is called ‘binning’.

RGB2Gray Method

When obtaining the image from the music discs, white light sources are used
for both 2D and 3D imaging. Specifically, the red light is positioned perpendicular to the disc surface in place of the white light. The blue light and the green light, the other two of
the three primary colors of light, are positioned at 45 degrees to the
surface of the disc surface. Therefore, the processed image of the groove will
have three colors: red, blue, and green. The edge of the disc could be
identified by one of the colors. When wanting to see only one of the three
colors from the image and excluding the rest, we can use the function
‘RGB2Gray’ in the weaver project. We can set the desired color weights, or the
proportion of the color, of red, blue, and green in the image.

However, the weaver project right now has one disadvantage of tracking by
color: tracks by different colors will start with different points in the
audio, making the final processed audio, which will add the three audios
together, less coherent. I will further investigate this, automating the
process of running the weight of the different color to process the image and
improve the resulting audio.

Difficulties faced, and solutions

To make the development smoother, compiling the code to generate a
binary is necessary. I will further investigate this along with my
mentor.

Topics of next week

Here are some topics that might appear in my next week’s post.

1. Introduction to the function ‘ShiftTrack’ and its example.
2. Introduction to the Fast Fourier Transform (FFT), which can help engineers measure the noise in the audio.

Thank you, everyone! I am so excited about this project and am looking forward to
your comments/suggestions!

2 Replies to “Week 1: Starting the Weaver project, loading image, etc”

  1. Monali G. says:

    Wow Leo, this is impressive! I can’t wait to hear more about compiling your code.

  2. Eric M. says:

    Sounds complicated. But it seems like you’ve begun the interesting parts of the internship. I look forward to what comes next!

Leave a Reply