Sample Manipulation 4 – Databending

I started talking about manipulation of samples by describing The Black Widow, a PureData patch for controlling up to 4 samples using a flight simulator-type joystick control, and the Black Widow II, with improved features and an automatic mode.  I also described two other applications by Karlheinz Essl and Kevin Holland (Sineqube) which work similarly, manipulating 4 samples in real time.  The third post in the series described two applications which work in real time to manipulate a single, longer sample.

This post covers some non-real time methods of working on samples, often known as ‘Databending’.

The basic idea of databending is very simple: open a sound file with an application it’s not meant to be opened with, change the file in a semi-random way, then save it.  When you play it again as a sound file, it will sound different from the way it sounded originally.  Often quite a lot different.  Some good advice on how to do this can be found at http://www.intelligentmachinery.net/?page_id=29.


First of all, what sort of files does it work on?  I’d say you should always start by trying different formats of the same sound file – certainly an uncompressed format like aiff or wav, and a compressed format like mp3 or flac.  There can be a big difference in the results.  Once you’ve tried a particular technique on two or more formats, you might decide it generally works better on one of them, and then concentrate on that one; the next technique you try might work more effectively on another type.


Secondly, try some files with different musical content: one file with, say, fast percussive content, another with slow, sustained passages.  A particular technique might suit one style, but not another.


Next, what ‘other’ applications could you try?

Text applications are always good.  All files can be displayed as text, as you will have noticed if you’ve ever had occasion to read one – trying to open a pdf in Word by mistake, for example.  It generally looks like gobbledygook most of the the way through, although there’s occasionally a ‘header’ at the beginning with some readable stuff in it.

Sometimes, even the mere act of opening a sound file in a text program and immediately saving it can make significant changes to the way it sounds when you play it back again.  Notepad in Windows or Text Edit on a Mac can be used in this way.  Provided you save the file as ‘text’ and keep the extension (e.g. ‘.wav’ or ‘.mp3’) the same, you’ll be able to play it back.

The first example shows this.  The first extract is from the original sound file I used; the second extract is the file opened and saved in Notepad on a PC; the third extract is the file opened and saved in Wordpad.

Once you’ve got the file in a text editing application, however, there are many things you can do with it.  I’ve tried the following and got some interesting results: changing all lower case letters to uppercase; changing all zeroes to 1’s; changing 1’s to zeroes; cutting and pasting text from other places into the middle of the file; and swapping bits of the file around.  Occasionally you’ll end up with a file that won’t play at all, occasionally a file that sounds exactly the same as it did before – which is particularly disappointing – but usually you’ll have changed something about it in a hopefully interesting way.

Incidentally, I normally use ‘Save As . . .’ rather than ‘Save’, and give the new file a name which reminds me what I’ve done to it.  This has the double advantage of helping me to achieve the same effect again by working out what I did to the file, and ensuring that I don’t lose an original which I might need again – either in its original form, or as the source for more experimentation.


Some of these files that you end up with can be rather unstable and may sound different depending on which application you choose to play them back in.  So my next piece of advice would be to try some different applications to audition the results of your work – Windows Media Player, Audacity, VLC, Quicktime, and so on.

If you have a way of re-recording them as you play them, this would be good, as you can then save a stable file which will sound the same whenever you use it.

The second example illustrates this.  The first extract is the original sound file I used; the second extract is this file after processing in BBEdit Lite, played back using Videolan VLC; the third extract is exactly the same BBEdit Lite file played back using Apple Quicktime, producing a significantly different result.

Both files were re-recorded to preserve these differences permanently, as described above.  I use an excellent application called Wiretap Studio for this, but it isn’t freeware.  There are ways to do it without having to buy extra software – free applications like Soundflower (which I think only exists for Mac) or Jack will enable you to route the output of one application, which plays the sound file, to the input of another (like Audacity, for example), which records it.


Another, less obvious, way to ‘bend’ sound file data is to open the file in a picture editor like Photoshop, apply an effect and save the file.

This is slightly more complicated; I don’t know about other applications, but Photoshop won’t recognise sound files as pictures, so you have to do a couple of things to persuade the application to try opening it.

First of all, make sure to choose ‘All Documents’ (as opposed to ‘All Readable Documents’) from the ‘Enable’ list.  This makes all files, even sound files, potentially openable.

In my experience the ‘Format’ option was greyed out, but make sure it shows ‘Raw’ to enable a dialogue box like this to open up:

Clicking ‘OK’ should enable Photoshop to open the sound file, and give you a visual representation of it.  It’ll look something like this:

If it doesn’t open and you get the dialogue box at the top of the following picture – which happened to me a lot – reduce the numbers indicated until you get the dialogue box at the bottom, then it will open.

After that, you can apply some effects to the file, then save it.  As mentioned above, I would keep the suffix – usually .aif or .mp3 – the same, but change the file name to reflect the process it had been subjected to.  I had some success with Gaussian Blur, Noise and Despeckle, but any one of the many effects could do something to the sound.

The first example here shows the original sound file; this file with a Gaussian Blur filter applied; then this file with a Despeckle filter applied.

The second example shows an original .aif file with a Noise filter applied; and an original .mp3 file with a Noise filter applied, to show the different effect an identical filter may have on a sample which was identical in every way except the file format.


If you have problems opening a sound file saved by a graphics application such as Photoshop, the free Audacity can probably help.  Even if you get the dialogue box shown at the top of the following picture, you can still use ‘Import’/’Raw Data…’ to open the file, and then save it in the format of your choice.

Using the above methods you should be able to make some random, unpredictable and interesting changes to your sound files.


0 Responses to “Sample Manipulation 4 – Databending”

  1. Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


June 2017
« May    

Enter your email address to follow this blog and receive notifications of new posts by email.

%d bloggers like this: