If you are new to the concept of using an audio editor like Audacity as a video processing tool, here’s the insta-crash course. Many online sources on this technique are full of inaccuracies and references to older versions of the app, so here I’ll provide a handy update on data bending video in an audio application.
In Audacity, you can Import any kind of file as Raw Data and then apply its audio effects processors to the data to creatively alter the data that makes up the video file, so that it’s glitched when reconstituted. However, there’s a bit of preparatory work to do first, and some pitfalls to watch our for. In honor of the upcoming US election, let’s glitch a video of the American flag which can be downloaded from Pixabay’s royalty-free video area. This will download as an .mp4 file.
The first step is to convert this .mp4 file into an .avi file. Here I’ll use the free version of Prism Video File Converter to make this .mp4 of an American flag flapping in the wind into an .avi file.
Convert your .mp4 video to an .avi file.
Next, we need to ‘trick’ Audacity into letting us open a video file inside it. To do this, we just Import the .avi file as Raw Data.
Import the .avi video file as Raw Data
Choose A-Law or U-Law as the encoding type. If you play around with the other settings besides this mandatory encoding one, you’ll get different results though these may not be very noticeable, given the unpredictable nature of data bending. Since with glitch effects there’s already a high degree of uncertainty involved, I tend to leave these settings as they appear and only alter them after I’ve developed a better sense as to what glitch patterns will be generated by audio effects in Audacity.
Pick A- or U-Law as the encoding setting. For the other settings, whatever is the name of the game.
A video file will generally resemble noise as a waveform.
Video file data inside an audio editor (Audacity) as raw data.
Pretty much all the online sources say to apply your audio processing to the areas of the file between the two end points, where important header and footer information is stored to keep the overall file type integrity intact. In between these end points, however, you can go wild supposedly, as per all the inaccurate online sources.
Below, some low pass EQ, wahwah, echo and distortion is applied to the main segment of the file between the end points. Note that we are not processing the start and end of the file, taking care to select the bulk of the middle area. Also, various sources warn not to use effects that have the effect of changing the file’s duration, though I’ve found in my own experience that sometimes these work just fine despite all the online warnings about them. You’ll just have to experiment and see what works in your context.
Trying out some audio effects on our American flag data.
To export back out to video, make sure you Select-All on the entire file so that Audacity doesn’t try to render only the portion you had selected for audio processing, as those end points are needed to get a valid video file. Your Export option is Selected Audio.
Setting up your export to get back into video data space.
Export the file using the same encoding type as when you brought the file in, A-Law in this case. Also, this will be a .raw file by default, however you need to change the file extension to .avi either at this stage or after you save the file.
Export with the same encoding method, and change the file extension to .avi either now or after you export.
Now, your .avi file will make the same but reverse trip through your video file converter to be changed into a usable .mp4 format.
Convert now back to .mp4.
What often happens, if you follow the advice of the many online sources as I’ve done so far, is that your video will stop playing after the first few glitches.
Sigh, all that work wasted for an incomplete glitched-out video.
After some experiments I noticed something about the difference between videos I download from the internet, and videos I export from Davinci Resolve. Here is what the exact same flag video looks like when imported into Resolve, exported from it as the default H.264 .mov file type, then brought into Audacity in the same manner as described above.
Intermittent data structuring between the two end points.
This is quite interesting because it shows that, despite what a lot of online sources claim, there is more to a media file’s data structure than just the start and end points which are supposed to define the zone of free play data moshing. Each of these rhythmic nodes in the Resolve-exported file are key to the structural integrity of the file, which means that data bending needs to take place between each one of those nodes, and not just between the start and end points of the whole file as is so often suggested in online sources. I’ve verified this in previous tests where applying audio effects across the nodes wrecks the file, but you’re welcome to try for yourself! So, let’s stay within the node boundaries for this new round of trials. Here, we’ll just add in some distortion between each data node.
Data bending between the structural data nodes only.
When data bending works, you get so excited that you introduce a huge heading in a blog to celebrate. Whew hoo! What a difference a codec makes.
Bolt those mailboxes to the ground, you crazy American virus-loving yahoos.
Now your glitchy video is fully intact end-to-end. You can further refine the flow of the glitches through various postproduction processes in your video editor. Experimenting with blending modes to combine the unprocessed and glitched versions of the video clip is often a great way to preserve fluidity of motion and still get the full glitched effect.
In general, I recommend trying one effect at a time. This way you can verify that you’re maintaining overall file integrity, then pile on additional effects in between each data structural node (I think they’re called ‘rhythmic buffer packets’ or something, who cares : ) as you gradually ‘complexify’ the effect (yes, that’s a Scrabble word).
Now let’s apply different effects to different node-defined areas, in a rhythmic way.
Rhythmically alternating wahwah, low pass filter and echo across the data nodes. What will this actually look like? We shall see!
Yep, still glitchy, but not that different.
This combination of three effects isn’t having as profound a difference as one might hope for. Oh well, that’s glitch technique for you. In audio, when all else fails, pour reverb on everything and see if plunging the mix into a cavern can save it. So, here’s reverb to the rescue, applied in huge amounts to each safely editable area between the nodes.
Reverb will always save us.
Now it would be time for a drum roll, to build anticipation for the final video in this blog, only you’d never hear each drum hit in the depths of this reverb canyon we’ve doused our data in.
Reverb. It just works.
And so we complete our crash course on using Audacity as a video effects processor with an apropos glitchy portrait of America today. Don’t forget to vote, even if your ballot might be glitched by a populist mail carrier or ideological Supreme Court!
I use this technique in the music video below, where it shows up around three quarters into the video.