In my new G5 Mac, iTunes has a feature called the “Visualizer,” which is an amazing thing that produces sort of psychedelic visuals to accompany the music. It basically has nothing to do with the music, except that it does respond to volume - you can see the beat.
My first question is: how does it work, and how much memory is being used? Second question: how long can it play before repeating, if ever?
You can see the beat, but you think it doesn’t have anything to do with the music?
Unless the visualizer is very different from anything else that goes by that name in any other player I’ve run into, it’s entirely music-driven. Different parts respond to different frequencies–some low, some high.
No, you can change tracks into a completely different type of music, and the designs continue as before. And you can repeat a track, yet the designs don’t repeat.
Watch for a particularly jagged part of the display–sometimes it’ll be a circle or two, sometimes it’ll be a line or two. That’s the waveform of what you’re listening to, and that’s what drives the display. If you have trouble seeing it, try watching the opening movement of Beethoven’s fifth.
The effects are chosen at random, and are very likely based on a small number of variables with a few possible values for each. From there, it’s the power of combinatorics that drives the seemingly endless variety.
As an aside, this looks almost identical to a Winamp plugin that we used to watch in college. This would’ve been '99 or so. Did iTunes just add this recently?
Oh, on my PC, playing with the visualizer on takes up about 60MB of RAM, as opposed to 48MB from just having iTunes open.
I can’t say anything about the current set of visualizers, but I can explain how the one worked that I wrote on my first computer back in the early 80’s. It was a very simple concept and I was amazed at how well it worked:
At each instant in time, there is a number associated with the voltage on the digital to analog converter that ultimately drives the speakers. My program had a predefined number of shapes, colors and sizes that would get placed on the screen based on the numeric value of the voltage (the voltage really maps to the magnitude of the sound being created by the speaker at that instant). The more often the number was above a certain point, and the higher the number, caused more shapes and they lasted longer and changed more colors, etc. This was coupled with a pseudo-random number sequence to determine which shape specifically, how big, exactly which color, etc.
Despite not having any actual knowledge of the music, or any algorithms to attempt to see patterns in the numbers, it really seemed to create visual displays that “went” with the music.
As far as I know, iTunes has always had this feature. I bought my iMac in 2001, just after Apple released OS X, and I amused myself for hours watching the Visualizer. I don’t think iTunes existed before OS X, but I could be wrong.
Point of Order: It’s not strictly a Mac feature, it’s an I-Tunes feature. It’s running on my Toshiba right now.
I like to think of it as the Random Oskar Fischinger Generator.
iTunes may not have existed for MacOS 9 at a time when it didn’t also exist for MacOS X (i.e., Apple may have released an OS 9 and OS X version simultaneously), but there was definitely an OS 9 version.
I’d like to see a visualizer that does a better job of mapping sound to visuals. Like analyzing the sound and determining what key best describes the “chord” of sound being produced at any moment and assigning a color or shape to each key; and mapping one lineshape to the topmost frequency over a certain percent of total output volume, and another to the bottom-most, so they “trace” their respective pitches the way most visualizers “trace” the amplitude (volume); and some way of characterizing the quality (type, not goodness) of the sound, e.g., a middle C on an oboe should look different than the same middle C played at equal volume on a violin.
Be particularly nice if it were a stand-alone app that would simply map all sound going to the computer speaker, rather than being a feature of a given audio player, so you could use it with the audio player of your choice.
It could be done for MIDI files. The MIDI file format is basically an electronic representation of a score, and the player translates that into sound.
.wav and .mp3 formats, on the other hand, pretty much contain instructions for how the speakers should vibrate. Reverse-engineering chords and instruments out of that is intractable if not actually impossible.
FWIW, there are thrid party iTunes visualizer plug-ins available
How can it be impossible if humans can do it?