In 1999, Aphex Twin released a B-side track on the Windowlicker EP with one of the longest titles in electronic music history: a string of mathematical notation reading roughly "ΔMᵢ⁻¹=−α∑..." that fans took to calling "Equation" or "Formula." For two years it was just an abrasive five-and-a-half-minute squall of digital noise. Then in 2001, a fan named Jarmo Niinisalo dropped the track into a spectrogram analyzer and found something at the 5:27 mark: a grinning, demonic face staring out of the audio. He tweaked the scale to logarithmic and the demon resolved into Richard D. James himself, hidden inside the song.
This was the moment spectrogram art entered the public conversation. What had been an obscure technique known only to a few electronic-music producers became a music-history phenomenon: musicians using audio software to encode a hidden image in audio that would only be revealed if a curious listener thought to visualize the frequency spectrum of the song. The Aphex Twin spectrogram became the canonical reference point, the discovery that taught a generation of fans to look for a hidden image in audio recordings the same way they might look for a backmasked vocal in a 1970s rock track. This guide is the complete walkthrough of the canonical examples, how the technique actually works, how to decode a hidden spectrogram message, why audio steganography matters beyond easter eggs, and the tools you can use to make your own.
Aphex Twin's "Equation" - The Founding Moment of Spectrogram Art
The track that started it all: "ΔMᵢ⁻¹=−α∑Dᵢ[η][∑Fjᵢ[η−1]+Fextᵢ[η⁻¹]]," the B-side of the 1999 Windowlicker EP. Richard D. James used an early Mac program called MetaSynth, which converts images into sound by treating an image's pixel grid as a frequency-time map. He dropped his own face into the program, generated the resulting noise, and tucked the audio into the final 30 seconds of an otherwise unrelenting noise track.
The discovery in 2001 by Jarmo Niinisalo was the catalyst. Niinisalo published the spectrogram image on his bastwood site, and from there it spread to electronic-music message boards, then to mainstream music journalism. Vice, Mixmag, and Far Out Magazine all eventually covered the easter egg. The track became a kind of canonical reference point: any time a hidden image showed up in audio thereafter, it was being measured against Aphex Twin's face in the spectrogram.
The technique was not new (audio engineers had been able to do this since the 1950s with analog tape and oscilloscopes), but the specific cultural moment of an electronic-music artist hiding their identity inside a track was. It established spectrogram art as a recognizable form.
The Canonical Examples
Five spectrogram easter eggs that show up in nearly every history of the form, all verified with primary or credible secondary sources.
Aphex Twin - "Equation" (Windowlicker EP, 1999)
Already covered above. The originator of the modern Aphex Twin spectrogram tradition. The face appears at 5:27 in the last 10 seconds of the track and requires a logarithmic spectrogram scale to render fully. The Aphex Twin spectrogram is still the most-cited example in any retrospective on the form. Far Out Magazine has the full history, and Vice covered it as the gold-standard easter egg in electronic music.
Plaid - "3recurring" (Rest Proof Clockwork, 1999)
Released the same year as Aphex Twin's hidden face, Plaid's track "3recurring" from Rest Proof Clockwork carries a continuous mural of the digit "3" running through the spectrogram. The image links visually back to the cover art of their previous album Not For Threes, and to the track title itself. The technique is a kind of visual cross-reference between two albums, encoded in the audio of the second. Mixmag's history of spectrogram art places this alongside the Aphex Twin example as a foundational case.
Venetian Snares - "Look" (Songs About My Cats, 2001)
The album title is the clue. Venetian Snares (Aaron Funk) included a closing track called "Look" on his 2001 release Songs About My Cats. The track is described in reviews as "very dense, noisy, and unlistenable" until you spectrogram it, at which point images of his cats render in the frequency spectrum. The album literally is what it says it is: songs about his cats, where the cats are visible in the audio rather than referenced lyrically. The Tumblr archive @spectrograms documented the full track, noting the cats are still visible even in 192 kbps MP3 versions of the track.
Nine Inch Nails - Year Zero ARG (2007)
Trent Reznor's 2007 alternate-reality-game (ARG) campaign for the album Year Zero used spectrogram art as a clue-delivery mechanism. A USB flash drive was found in a bathroom stall at NIN's Lisbon, Portugal concert containing a high-quality MP3 of "My Violent Heart." Listeners who analyzed the file with a spectrogram found "The Presence" - the godlike hand from the album artwork - encoded into a few seconds of static at the end of the file. On the album-version track "The Warning," the same hidden image appears at the end. The NIN wiki documents the full ARG timeline and the role spectrogram analysis played in unlocking the puzzle.
This was the first major use of spectrogram steganography as a narrative storytelling device rather than an artist easter egg. The audio carried plot-relevant imagery; finding it was part of the game.
DOOM (2016 OST) - The Cyberdemon Hidden Message
By 2016, the technique had moved beyond electronic music entirely. The DOOM 2016 game soundtrack composed by Mick Gordon contained a spectrogram easter egg in the Cyberdemon track: a hidden spectrogram message rendered in the audio's frequency spectrum that fans uncovered by running the OST through Audacity. The technique that started with Aphex Twin in 1999 was, by 2016, being used in AAA video game soundtracks as a hat-tip to the music-history precedent. Whatever spectrogram message you encode, the canonical pattern is consistent: the audio sounds like noise, the spectrogram reveals an image or text the artist intended you to find.
For the more comprehensive list of documented spectrogram easter eggs across decades, TwistedSifter compiled 11 examples including some from rock and pop tracks beyond the electronic-music canon.
How Spectrograms Actually Work
Spectrograms are visual representations of audio. Time is the horizontal axis, frequency is the vertical axis, and brightness or color shows amplitude (how loud each frequency is at each moment). When you record a piano playing middle C, the spectrogram shows a horizontal line at the frequency of middle C for as long as the note sustains. When you record a chord, you see multiple parallel lines for each note's frequency. When you record speech, you see complex patterns of bright spots that, with practice, can be read directly as phonemes.
Spectrogram art works by inverting this: instead of analyzing existing audio to see its frequency content, you start with a target image and generate audio whose frequency content matches the image's pixel grid. You feed the program a picture of a face; the program generates noise that, when re-spectrogrammed, renders that face in the frequency spectrum.
The audio sounds like noise to the human ear because the frequencies are not arranged in any musical pattern; they exist purely to draw the image when visualized. This is why every canonical example above is described as harsh, dense, or noisy when listened to: you are hearing the picture, played back as raw frequency content.
Why Audio Steganography Matters Beyond Easter Eggs
The technique extends beyond music-history easter eggs into legitimate engineering uses.
ARG storytelling. As Year Zero demonstrated, hiding plot-relevant imagery in audio is a way to deliver clues to audiences willing to analyze the source files. Several modern ARGs continue to use the technique.
Forensic audio watermarking. Audio professionals embed inaudible watermarks into music releases for theft tracking. If a track leaks, the watermark identifies which pre-release copy it came from. This is a related but distinct technique from spectrogram art (the watermark does not need to render as a visual image), but the underlying principle is the same: encode information into audio in a way that does not affect human listening but can be extracted by analysis. For a creator-focused walkthrough of forensic audio watermarking specifically, ProveAudio's audio watermarking platform covers the workflow.
Archive preservation. Library and museum archives sometimes embed identifying tones in audio digitization workflows so that misfiled or copied files can be traced back to their archival source.
Album art puzzles. Beyond ARGs, some artists use spectrogram art as a layer of album-art interactivity, rewarding fans who explore the audio with images that complement the visible artwork.
The common thread: audio is denser than human listening can extract on its own. Steganographic techniques use that density to carry additional information beyond what listeners hear directly.
How to Make Your Own Spectrogram Art
The conceptual workflow is consistent across tools:
- Pick a target image (usually high-contrast, simple shapes work best because the spectrogram resolution does not preserve fine detail).
- Run the image through an image-to-spectrogram converter that generates audio matching the image's pixel grid.
- The output is a noisy audio file that, when played, sounds harsh.
- Open the audio file in a spectrogram analyzer (or share it with someone who has one), and the original image renders in the frequency spectrum.
Several tools handle this conversion. The historical canon includes MetaSynth (Mac, the tool Aphex Twin used in 1999, still actively developed), Coagula (free Windows tool with a small but devoted user base), and ARSS - The Analysis & Resynthesis Sound Spectrograph (cross-platform, open source). For people who want the spectrogram-art workflow without learning a new audio program, the consumer-facing option is Img2Sound, which compresses the four-step workflow into upload-image, download-audio, with no audio-engineering background required.
For musicians and audio engineers already working in a DAW (digital audio workstation), Audacity has a built-in spectrogram view that lets you visualize any audio file's frequency content; it does not generate spectrogram art directly, but combined with a tool like Coagula it covers the full round-trip workflow. SonicVisualiser is the more advanced free option used by audio researchers.
For music creators interested in the broader audio-tools landscape (watermarking, stem separation, transcription, spectrogram art), the Luxaris Digital roundup of audio tools for musicians, producers, and podcasters covers the workflow categories side-by-side. For the question of which AI tools generally are worth the spend at the consumer level, Built With Our Own Tools is the operator-perspective companion piece.
A Note on Difficulty and Time
Spectrogram art is technically simple but creatively hard. The technical step (image goes in, audio comes out) takes minutes once you have a tool installed. The creative step (picking an image that will render legibly given the frequency-resolution constraints, deciding where in your track to hide it, deciding whether the harsh noise should be loud and surface-level or buried in a quieter passage where listeners might miss it entirely) takes considerably longer.
The Aphex Twin example works partly because the underlying track is already abrasive noise; the spectrogram-art passage at 5:27 does not stand out from the surrounding 5:27 of equally-abrasive material. The Year Zero hidden hand works because Trent Reznor framed it as a deliberate ARG clue that fans were invited to find. The choices about how visible to make the easter egg are part of the artistic decision.
For first-time experimenters, low-stakes is the right approach: pick a simple symbol or text, hide it in a side-project track, see if anyone finds it. The technique scales from "fun easter egg" to "career-defining ARG" and the choice is yours.
Try It Yourself
The spectrogram-art tradition started with Aphex Twin in 1999 and has continued through Plaid, Venetian Snares, NIN's Year Zero ARG, and into modern game soundtracks like DOOM 2016. The barrier to entry has dropped dramatically: what required MetaSynth and Mac audio-engineering expertise in 1999 can be done in a single web-based step today.
If you want to try the Aphex Twin trick yourself without learning Audacity, Img2Sound handles the image-to-audio conversion in one upload. Pick an image, get an audio file, share it with the people who will know to spectrogram it. You will have produced your own contribution to the spectrogram-art canon, joining a tradition that is now older than most listeners.