1

I would like to ask on what approaches could one attempt to generate a colored waveform out of an audio file. Below are two examples of waveform coloring by commercial products along an audio sample of that music at this position:

Traktor Pro

enter image description here

Pioneer Rekordbox

enter image description here

Audio sample at this position:

https://youtu.be/42t0Sdy-NUw?t=96

By looking at the output, one can infer the following:

  • low frequency content is colored in red
  • mid frequency content is colored in green
  • high frequency content is colored in blue

Obviously, this is a very simplistic analysis of the situation and doesn't really tell how it's been done.

Question:

What kind of algorithm is at work under the hood to produce such colored waveforms?

Disclaimer:

I have asked a similar question a while ago, now I feel it's time to ask it again for these reasons:

  • question has been asked 6 years ago!
  • the EchoNest web service used for detecting audio features has closed shop
  • couldn't really make good use of @pichenettes answer by lack of knowledge
  • still little public information on the web about this subject
  • there might be fresh minds that can suggest interesting approaches

(let me know how this question can be improved to be complementary to the one previously asked)

Eric Cartman
  • 284
  • 1
  • 8
  • my waveform coloring is just density: https://dsp.stackexchange.com/questions/184/calculating-the-pdf-of-a-waveform-from-its-samples https://github.com/endolith/scopeplot freesound's is density and spectral centroid: https://github.com/endolith/freesound-thumbnailer https://github.com/MTG/freesound/blob/master/utils/audioprocessing/processing.py https://github.com/MTG/freesound/blob/master/utils/audioprocessing/wav2png.py – endolith Jun 11 '19 at 20:11
  • I always wanted to try coloring them with an audio spectrum → color spectrum mapping, so that white noise is white, red noise is red, pink noise is pink, violet noise is violet, and tones of different frequencies show up as different colors, but audio frequencies are logarithmic and mixing to produce white requires linear spacing. :/ – endolith Jun 11 '19 at 20:14
  • thank you, that's very interesting and I'm looking at it :) – Eric Cartman Jun 11 '19 at 23:10

0 Answers0