In grahpics-design, vectorized content are much easier to work with version-control rather than raster images: much smaller space so more fit to Git -style version-controlling. Can I do it somehow with audio?

Helper questions

  1. Does there exist something like vector-formed -music or more generically vector-formed audio?

  2. Does there exist something like Vectormagic i.e. a conversion of audio from "raster" into vector -form?

  3. I want to do vector -form music to keep repos in a small space, does something like that exist?

  4. If audio in vector-form exists (which I believe), which format would you recommend for hobbyist activity such as sound-effects?

  • Are Bezier -curves used in audio -production, fine-tuning or in formats? I am trying to find more mathematically oriented -answers, there are surely something...
    – hhh
    Commented Aug 16, 2012 at 15:47
  • 1
    Note that "vector graphics" isn't actually a great description of "non-raster graphics", though the name is established. Lots of stuff can be described mathematically as vectors, including any rasterised picture or discretely sampled audio file. So it's not a surprise that the audio analogues to vector graphics aren't called anything with "vector" in it.
    – leftaroundabout
    Commented Aug 17, 2012 at 0:31

2 Answers 2


The closest equivalent I know of would be a Standard MIDI File (Wikipedia).

Like a vector files contain a description of a shape, a midi file contains a description of the tracks(layers), instruments, notes and timing. The file size is much smaller than the recorded audio and easier to edit.

Another text based representation of music is MML Music Macro Language a form of ABC Notation which was used for storing music on 8-bit and 16-bit game consoles where space on cartridges was a premium.

  • ...is Standard MIDI file the same as the thing you get out of most electronic instruments? Many different formats, which one is de-facto? Could you elaborate on MIDI -formats?
    – hhh
    Commented Aug 16, 2012 at 14:21
  • 2
    MIDI is fairly standardized; different implementations have certain quirks (like keyboards that think C3 is middle C instead of C4) but it's mostly standard. A MIDI file is essentially the same signals as you would get from a piece of hardware, but put recorded down with a timeline and occasionally meter information.
    – NReilingh
    Commented Aug 16, 2012 at 16:07
  • Also keep in mind that MIDI only captures a subset of a lot of things; it assumes a lot about your note scale, doesn't allow for much in the way of nuance, and so on. While it's a reasonable analogy for "vector graphics" it's not really that expressive for a lot of musical projects.
    – fluffy
    Commented Sep 1, 2012 at 21:51
  • I really like the way the MIDI files work.
    – user19338
    Commented Apr 11, 2020 at 11:22
  • I found this interesting paper regarding the midi approach by converting waves (raster) to midi (vector format) thru genetic algorithms. Seems to be quite expensive in terms of computation but it sounds very interesting as a starting point. researchgate.net/publication/…
    – abrittaf
    Commented Feb 20, 2022 at 18:09

I've wondered this too! In theory, splines can certainly be used to represent an audio waveform, and for certain signals, would be very efficient. They would not be efficient in general though, and they certainly won't make arbitrary audio signals fit any better into a text-based version control system like Git.

In digital graphics, vector formats are efficient at representing simple shapes and solid colours, but terrible at representing arbitrary bitmaps sampled from the real world through a camera sensor. In digital audio, the same limitation would apply: They would be good at representing simple-shaped signals, but terrible for representing arbitrary signals sampled from the real world from a transducer.

One useful application of vector graphics is to create precise, sharp shapes and colours (such as a logo), even if the end result will be rendered to a raster format. Similarly, vector audio may be a useful tool for synthesizing precise, mathematically-described sounds, even if they are immediately rendered to PCM for further processing.

A quick google search turned up this interesting thread on gearslutz where someone has described this idea with some diagrams.

  • 1
    While interesting, I'm pretty sure this isn't actually answering the question.
    – Warrior Bob
    Commented Aug 24, 2012 at 16:08
  • 1
    Maybe it was a bit rambling, but I did turn up the only external link describing a way to represent audio with splines. I think I answered the OP's 1st and 3rd question.
    – lukecyca
    Commented Sep 6, 2012 at 18:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.