The Internet has recently been populated by cool glitchy images made with machine learning programs, which create images based on user input.

If you haven't messed around with these already, I highly recomended it. This is the result of the prompt “Article about Databending with Audacity” fed into the starryai.com 'Altair' AI.

But did you know that you can also use audio editing software to create digital art? It is called databending, and the product of databending images is often called glitch art.

There are many ways to databend images, but for this tutorial, we are going to use the audio editing software Audacity. This software is available for free download on the Audacity website.

Getting Started

First, we need make sure that the picture in question is a format that Audacity can read. Audacity will not accept traditional image file formats, such as .jpg and .png, but there are several file formats that we can convert our picture to for Audacity to read, namely, .tiff and .bmp.

For this tutorial we will use a .bmp image format. There are many free online converters available to convert your image from its present format to the .bmp format.

If you want to follow along without selecting or converting an image of your own, I've gone ahead and converted this one to .bmp, which you can download directly by right clicking and selecting 'Save Image As'.

Anyone who has used a computer with Windows XP will recognize this classic. Fun fact: this image, titled Bliss by photographer Charles O'rear, is likely the most viewed photgraph in human history. It is estimated to have been viewed billions of times.

Anyways, now that we've got our image in the proper format, we can get this imported into Audacity and start databending.

Importing the Image

Once we have Audacity open we want to go to File > select Import > then select Raw Data.

Next, Audacity will prompt us with import options. Be sure to select U-Law or A-Law for Encoding, and remember which option you selected, because you will need to export it using this encoding as well.

Now we should be able to see our imported "audio" file.

What you are seeing here (if you don't know already) is a waveform. Waveforms are visual representations of some sort of signal. In music production, it essentially represents the amplitude (or volume) of the audio signal over its length. In this case, we are seeing a visual representation of the raw data of our picture.

Pressing play makes for some really weird sounds. I do not recommended doing so at high volumes unless you happen to be a big Merzbow fan (Disclaimer: extremely loud).

Adding Effects

Now that we have the waveform of our picture in Audacity, we can apply effects to different sections of the waveform to manipulate the data in different ways. To do this, we want to use the cursor tool to select a part of the waveform and then pick an effect to use.

You can select any effect you like, but some will make the file unreadable, and selecting any parts of the file at roughly 10 seconds or before will mess with the file header and cause it to be unreadable.

This point in the process is pure trial and error, but I recommend starting with the Compressor, Invert, Echo, Reverb, and Reverse. Fair warning, do not be discouraged if you do something that makes the file unreadable, this can happen when databending with Audacity. Just start over, and try again!

Once you've applied some effects, it will look generally the same as it did before, but with some minor changes to the waveform:

When you are satisfied that you have sufficiently messed with your waveform, it is time to export it and see what you end up with!

Exporting

There are 4 settings that need to be properly selected here, so be sure to follow carefully.

    1. You need to include .bmp at the end of your filename.
    2. For the dropdown option 'Save as Type' select Other uncompressed files.
    3. For the dropdown option 'Header' select RAW (header-less).
    4. For the dropdown option 'Encoding' make sure you select the encoding type you did before (U-Law or A-Law). In the case of this tutorial, we did U-Law.

Other than applying an effect that makes the file unreadable, this is the point in the process you will be most likely to make a mistake, so be sure to double check your settings each time before you export!

Conclusion

What we have done, essentially, is converted an image to a format that Audacity accepts as raw data. Then, we tricked Audacity into thinking we were giving it an audio file. In essence, it is all bits and bytes either way, so we can use traditional audio effects to manipulate the data of the file. Finally, we export the data back into the format we started with, but this time with some cool glitchy effects!

Now, we can check out the finished product in whatever folder we exported it to (spoiler: you already saw it in the article thumbnail).

Should we give it a name? Glitchy Bliss? Gliss? Bli-- forget it. Well, I hope you enjoyed this tutorial, thanks for reading! Reach out to me if you have any questions about the process, and remember that there is a lot of trial and error involved. It will take a few tries before you end up with a product you are happy with!

Once you get the hang of it, though, you can start to make some really interesting compositions. Have fun, and happy databending!