Noise school #666: abusing "input raw data" in Audacity

Hey, so been a while since I've posted, I've been beyond exhausted. However I have something simple, quick and fun you can try to make small harsh noise clips. (or entire songs if you are lucky and don't kill your computer. be careful!)

These small clips can be very useful for glitchy sound design and I frequently make use of this technique to get free unlimited noise bursts samples. Indeed you could even import such samples into a slicer and make a glitch-instrument with them. Often I blend these glitchy sounds into other sounds (like pedal noise) in a slicer, squishing them together with distortion and compression (the massive DC offsets can cause interesting things to happen here if you do this).

First, get Audacity

You should have it already by now. It's like the duct tape of music software. Go get it. I downloaded it without extras.

1: Import as Raw Data

Import the data. RAW. You know you want to.

Literally any file on your computer can be imported. However depending on the file type (and its size) results will vary. Wildly.

3: the import menu

First, changing these settings will change how Audacity 'interprets' the file.

Here are my 'default' settings.

Encoding: The various 'encoding' options can mean radically different sounds for the same file. Experiment with these! Especially good for files that usually just produce white noise (i.e. pngs).

Each has their own character too. GSM 6.0 sounds like an ultra-loud screeching robot. PCMs tend to produce a lot of white noise. NMS ADPCM tends to producer quieter 'squelchy' sounds with occasional random bursts, for example. This can vary file type to file type too.

Byte-order: These options don't seem to change the files too much, but could be wrong or just so far unlucky. Can't hurt to mess around with.

More info about this idea here: https://en.wikipedia.org/wiki/Endianness

Start Offset/Amount to Import: tells Audacity how many bytes 'in' to the file it will start importing. The higher it is, the shorter the sound will be.

'Amount to import' seems to have a similar function. Can be useful for extracting a small chunk from very big files. But I don't really use it for this purpose.

Sample rate: Same meaning as it usually does for sound, but here you don't have to pick 44.1/48 as usual. Lower sample rates can be useful for getting less "harsh" sounds and tend to sound like a sound effect from Wolfenstein 3D. Make sure you have a DAW that can handle low-quality sounds without breaking.

Detect: Here Audacity tries to guess the encoding itself, which is useful for importing audio files in strange formats. Here its irrelevant, as were are not importing audio. It's better to just pick and encoding type yourself and see which makes the "best" sounds.

4: why does it sound just like white noise?

Usually the first thing that people put in is an image file, such as a jpg or png. Because of how these are coded, they generally just make white noise.

So try these instead if you want cooler/sillier sounds:

PDFs (recommended!)
Applications (be careful! use ones you don't need)
HTML files
Video files (especially the first part of the file, the rest is white noise), but be careful with large files).
*some* word files (some are too small and only make milliseconds of sound).

You can also try changing the encoding type with the same file to see if you get more compelling sounds with a smaller entropy (White noise is a high/maximum entropy waveform)

--

Conclusion: Don't forget these sources don't come from natural sound, but pure binary code. Consequently there will be pretty big DC offsets and the like in the waveform which might distort the song they are imported into in various ways (often they are quieter, then with a click). If processing in Ableton, I'd recommend the Saturator with the "DC" button on.

For the same reason, there will be a LOT of frequency content and obviously it will be very loud. You can change this later, but just make sure you don't have your speakers too loud when trying for the first time.

WARNING: be careful about importing large files as while Audacity is reading them they will be temporary storage and thus will clog up space. You can chew through HDD space and can cause crashes. The cache should reset upon a computer restart. I'd recommend not using files over 1GB unless you have a lot of spare storage. That being said, you can generate song-length noise at very large file sizes... have fun but be careful as Violet (vi a./teeth dreams) crashed her computer by importing a video trying this method (the computer is fine now lol it didn't break).

Also, maybe don't import important applications and files (like your OS). I don't know if it will break it, but don't fuck around and maybe find out.

Here is a demonstration WAV: it's this post (as a .pages document). OooOOoo postmodern.

---

Update: Nova (pathological function) as added some useful notes about the above, I've pasted them here.

on sample rate: you use a loopback like blackhole (software - mac) or some usb interfaces that have loopback channels to record low bitrate output as industry standard bitrates in a different program

other tricks: for applications just copy it out of the applications folder into your home folder first if you're concerned to guarantee the original won't be effected. larger mac applications are good for this as they are basically fancy folders with a lot of varied data inside. larger zips, dmg/pkg files or isos might also be interesting as they are all collections of other data bundled together. compressed packages will probably be more white noisey.

fun kind of: you can use command line tools to look at files or commands as text and send it to a command line player - the mac one is called sox and you can get it from homebrew, linux has a built in one called aplay - but you can bypass using that by just saving the output as text. 

so commands like "ls -Rv ~" which lists every single file in your home folder as an example - the structure of the text makes it sound different to a normal document. you save it to text by typing something like "ls -Rv ~ > homefolder.txt" and then opening that file in audacity.

replace the ~ with a different folder like "/" for the whole computer or a specific folder.  it won't change anything on your drive as it's just listing all the files one by one in the command line, though it can run for a long time - control+C will force it to stop. because of the way files are structured you end up with a lot of lines that are very similar all in a row, so it can end up being more beat-repeaty than "normal" text.

corruption: you can junkify data by forcing wordpad* (the built in one not ms office word, not sure about non windows options) to open files it's not supposed to and saving them and going into audacity as it will insert a lot of random bs this way. for BMPs this tends to make them weird, so it might also corrupt wavs