|
|
|
|
|
|
by toemetoch
5145 days ago
|
|
|
If done properly (linear vs log scale, well-chosen cut-off frequency, thresholding values and discrete colors) you get a representation of the patterns in the song. With electronic/techno music you see where the intro/beats are. With other music you can see where the intro ends, the chorus repeats, instrumental solo, ... The way it's done now you see a bunch of peaks/vertical bars. Unless there's a change in the power of the song the bars stay pretty much at the same level. It's a lot of real-estate that can be put to good use. |
|
|
Absolutely not you don't see that, how can you see where anything starts or ends, when it's not a temporal representation? The only thing it represents is the presence (or lack thereof) of frequencies in a song. For example, what does this http://i.imgur.com/QRka5.png tell you about the song? Can you tell where the intro or chorus or whatever else begins or ends? A temporal representation works way better because you can skip to parts of a song. A spectrum can't let you skip to anything, nor can it give you any information that the general population would like to know about a track.