Spectrogram Art: A short history of musicians hiding visuals inside their tracks (Mixmag)

[EERecordist]

https://mixmag.net/feature/spectrogram-art-music-aphex-twin

No idea what it sounds like. Steganography.

In RF I usually see time on the vertical axis and frequency on the horizontal axis, usually called a waterfall.

I think some of the article is time horizontal and frequency vertical. If you are doing time horizontal, the bass and beat would be toward the bottom of the stream. Then you could manipulate the harmonics which people may not notice much in the music, especially related to the beat. Of course with noise music (a genre) you can probably do anything.

 

[kemmler3D]

Sometimes called spectral synthesis IIRC, this was hard to do when Aphex Twin made this, but has been easy for a while, FL Studio (Fruitloops) had a plugin called Beepmap to do it about 15 years ago. Basically this kind of software will interpret any image as a spectrogram and synthesize the tones it indicates, more or less how you describe above. Usually the color of the pixel determines volume, X is time and Y is frequency.

I actually went to the extent of drawing spectrograms by hand to produce certain sounds using this method of synthesis, and wrote an entire track using the results:

(I would not say I'm proud of it as music, but it does showcase the technique.)

If you use one of these tools on a simple image and place the sound in the song as-is, it will show up on a spectrogram like this.

Question: Do you think it counts as steganography even though the 'hidden' message is not really hidden? You can easily hear this stuff, it's just not clear what it is when you hear it.

 

[Ropeburn]

Basically this kind of software will interpret any image as a spectrogram and synthesize the tones it indicates, more or less how you describe above. Usually the color of the pixel determines volume, X is time and Y is frequency.

This is key to this sort of thing. You don't manipulate audio to create images in spectrograms, rather you generate audio from images. Important distinction and not intuitive, you have to think arsebackwards, so to speak. The software is basically a reverse spectral analyser.

A rather old but famous (in appropriate circles) example of software for this is Coagula:

Basically an image editor with painting tools and processing tuned to purpose, and image import. Plus the synthesis part of course.

Still available for free (x86), although I don't know about compatibility with newer OSes.

Coagula

 

[lucian]

Sometimes called spectral synthesis IIRC, this was hard to do when Aphex Twin made this, but has been easy for a while, FL Studio (Fruitloops) had a plugin called Beepmap to do it about 15 years ago. Basically this kind of software will interpret any image as a spectrogram and synthesize the tones it indicates, more or less how you describe above. Usually the color of the pixel determines volume, X is time and Y is frequency.

I actually went to the extent of drawing spectrograms by hand to produce certain sounds using this method of synthesis, and wrote an entire track using the results:

(I would not say I'm proud of it as music, but it does showcase the technique.)

If you use one of these tools on a simple image and place the sound in the song as-is, it will show up on a spectrogram like this.

Question: Do you think it counts as steganography even though the 'hidden' message is not really hidden? You can easily hear this stuff, it's just not clear what it is when you hear it.

i actually like it - you have more of it?

 

[kemmler3D]

i actually like it - you have more of it?

Well, thanks! There's a whole album, but only that track used the spectral synthesis technique. But stylistically the rest is not too far off.