First and foremost, There’s a web site called Fake Music Generator, at http://www.fakemusicgenerator.com/
Which uses a computer algorithm to dynamically generate albums complete with songs and cover art.
Here’s an example:
So today, I saw an article on Mental Floss which posted the ‘discovery’ that ‘musicians’ are embedding photos in the songs. Now it’s my belief that there’s much more than this getting embedded, computer programs, messages, and the like. you just gotta know what to look for and how to pull it out.
In any case. They listed an example., Where the music contains images of cats embedded in the spectral diagram – by a group they referred to on last.fm as “The Venetian Snares”, the title of the song – “A Song About My Cats”….
So I went to take a look at the band on last.fm, and lo and behold what do I see?
Shitpoor Artwork which mysteriously resembles the autogenerated art on the web site fakemusicgenerator.com, and weird song titles which have the same exact pattern.
And on top of that, bizarre looking titles of the songs with patterns once again resembling the same patterns as seen on fakemusicgenerator.com.
Now this is NOT to say that last.fm has legitimate music, which it most certainly does with Coldplay and the like. But my question is: how can Last.Fm distinguish content like this from the real gems? That’s a bitch of a question. I mean, I spotted the similarity immediately to the fakemusicgeneration, but how do you build in the likeness detection in an automated way?
And in any case. Has the internet gone collectively braindead?
First off, isn’t it obvious that the music was autogenerated with the weird nature to it?
Second, this ‘finding’ details HOW that autogeneration works, thus asserting my initial suspicions. By taking imagery, or set of imagery, translating the images from a spectral graph to chords, viola – you got a song.
As for the numbers on last.fm and #s of sales. I call bullshit. claims of popularity does nothing to describe actual popularity. You should have learned this lesson from Youtube.
In any case. The article asserted my suspicions that the autogeneration has something ‘underneath’ it, this merely validates it!
Now I want to know – where can I get ahold of this music generation?
And I suspect Hoillywood has one for movie generation too, with it’s ‘formulaic approach’
Now my question is – how do we ‘weed out’ autogenerated content from cluttering up and confusing legitimate sales channels? It creates noise and confusion.
That is: There MUST be a way to segment and reward users for reviewing autogenerated content that are segmented from traditional sales channels (not to say there’s not potential gems to be found) – this declutters the traditional sales channel and also allows us to ‘browse’ without spending all day browsing trying to figure out the crap from the creme.. .
Thank you, Mental Floss, for this find!
Have you ever listened to a song and felt like the music was painting a picture in your mind, or sending you a secret message? Oddly enough, that might not have been your brain playing tricks on you. Artists often include things like Morse code or manipulated sound bites in their music, adding interesting extra layers and producing “secret messages.” But some synthesizer artists, like Venetian Snares, use computer programs to convert images into sound—and then put those sounds in their songs. A track from one of his albums, “Songs About My Cats,” when analyzed with a computer program, produces sound waves that look like cats, like this:
This is a “spectrogram,” and it is a frequency (kHz)-time analysis of sound instead of the typical, spiky-looking decibel (dB)-time analyses you’re used to seeing on your music player’s equalizer.
Online images and videos of the spectrogram of Venetian Snares’ song “Look” abound, but other works by other artists—like Aphex Twin—are out there, too. Perhaps the most famous of spectrogram pictures is the “Demon Face,” found in Aphex Twin’s “[Formula].”
This picture isn’t actually a demon—it’s an image of Richard David James, the man behind Aphex Twin—but it’s still really creepy, and a perfect enhancement to the song (not to mention a gift to inquisitive fans).
You might wonder if these works of sound-art are photoshopped. You can test it yourself. We used two free programs—an image-audio encoder and an audio editor—and copies of both the song “Look” and a Microsoft Paint™ drawing of a smiley face.
We recorded a sound and analyzed it. We used a binaural beat (two sine wave sounds between 0 and 1000 Hz that are less than 30 Hz apart), which creates a vibrating, headache-inducing sensation in the ears, to get the image below.
Next, we imported Venetian Snares’ song “Look” and turned that into a spectrogram. Even with the quality of the free software, you can still make out what appear to be several cats’ heads.
But are we seeing the cats’ heads in those sound waves because we expect to see them? As a final test, we imported a smiley face from MS Paint into the image-sound encoder. This encoding software takes the image data and turns it into sound data.
After the program was done converting, we took the resulting .WAV sound file and imported that into the audio software, and analyzed it. Lo and behold! The unmanipulated result: Not a bad-looking smiley face, if we do say so ourselves.
Even with simple software, these proof-of-concept tests show that not only are the two well-known and highly circulated spectrograms of Venetian Snares and Aphex Twin songs real, but they can be encoded and decoded back and forth. And, anyone with a computer—and good anti-virus software (just a heads-up)—can make wildly-colored sound art. One tip, though: Don’t listen to the sound files—unless you just enjoy weird, dial-tone-like static sounds.