Before we start in Audacity, using the right image format is vital to minimise the possibility of an unreadable file. Uncompressed images typically respond much better to editing processes than compressed images, allowing you to be more flexible with just how much you edit the image in Audacity. In short, this means no JPEGs, PNGs or TIFFs.
Two formats stand out with the best results: Bitmap files and RAW files. Both have their advantages and disadvantages which also affect the final outcome after sonification.
Of the two formats, Bitmaps are the easier to read and use. There are fewer variables to set and the results can be viewed instantly after export. This is due to self-contained header information which specifies details about file size, dimensions, bit-rate and other information required by a computer to read it. This essentially means they are universally readable on any system and in default preview/viewing applications. The downside is that they aren't quite as flexible to certain changes as RAW files, however with the right care (explained in detail on this site), bitmaps are perfectly workable. For anybody who is new to the process, would like fewer steps or does not have experience working with headerless RAW, it's the best format to work with and the one I chose to use in all of my tests.
When exporting in Bitmap format, make sure to specify either 24 or 32 bit as 16 does not respond as well to editing in Audacity. All files on this website are 24 bit (8 bits per channel, 3 channels).
RAW files are typically very difficult to break when sonnified, even if only a fragment of the original code remains. This is due to a more flexible tolerance for missing bytes and the lack of a header to corrupt. Due to the absence of a header, image details such as dimensions and bit-rate must be remembered manually. A RAW file can also be exported with a header, in which case these details are automatically contained, however care must be taken to ensure that the headers is not edited during sonification in the same manner as a Bitmap. To complicate matters further, in built image viewers such as Photo Viewer on Windows and Preview on OSX are not always able to view these RAW files, requiring another third party application to open or re-save them. For these reasons reason, bitmap is the easier of the two formats to work with.
Selecting, Cutting and Pasting
Familiarise yourself with the Audacity interface first. Cutting and pasting is the first tool we can experiment with. Reorganising chunks will allow you to create a chopped up image. Care should be taken not to reposition or cut through the header. Furthermore, due to way Bitmaps interpret file size, the final track should not be shorter than the length of the original track. In the example, the track is just under 3 minutes long, therefore the edited track must not be shorter than 3 minutes or the file will register as corrupt and unviewable once exported.
Importing a Bitmap
Exporting as Bitmap
1 - 2
When importing Bitmap files, only U-LAW and A-LAW encoding can be used. Any other encode will result in a corrupt file upon export. Typically, more data is lost in the conversion using U-LAW, resulting in artefacting even before an effect is applied. A-LAW on the other hand converts the image without any artefacts. Preference is based on your intended visual outcome, however I have used A-LAW for greater clarity.
File>Import>Raw Data>*select file*
File>Export Audio>*File Extension to .bmp*
Locating the Header
1 - 2
In order to avoid total corruption of the file, the header must be located and deselected to ensure no effects are applied to it. To find the header of a Bitmap, send the cursor to the start of the track using the 'Skip to start' icon and zoom in until the waveform is represented as a series of dots. The header will most often always end with a long string of dots after which the safely editable contents will follow. Simply select all of the track past this point.
If you are importing more than one file, ensure that only one header is present per total tracks to avoid corruption. In this case, keep the header of track 1 and delete the header of any tracks that follow it. The final image will take the dimensions of track 1's header.