User talk:Moehre1992

Audio codec comparison spectrograms
Greetings. I'm sorry to disappoint, but the graph you added to many audio codec articles wasn't appropriate for most of the articles, so I removed them from all but audio compression (data). I think it can stay in that one article, as long as it has an appropriate caption.

The captions you used seem to indicate that you're under the mistaken impression that you can infer something interesting/alarming from the degree of lossiness reflected by the visual differences among the spectrograms.

The fact that the spectrograms of lossy encodings differ from those of the original audio is an indication that the lossy codecs are indeed lossy. However, it is a mistake to imply, or to even allow a reader to assume, that the "striking" differences in high-frequency content are indicative of anything whatsoever. Likewise, the visual differences between the lossy codec spectrograms doesn't reliably imply anything.

Frequencies above 16 kHz are quite often just noise and/or masked harmonics, inaudible to most listeners. Some codecs (like MP3) have less accuracy at higher frequencies, and most require more bandwidth in order to handle increasingly noise-like signals (as is often found up there), so the use of a lowpass filter actually increases the bandwidth available for encoding lower frequencies. In other words, the reader cannot be left to infer that a given lossy file is perceptibly better just because it has more high-frequency content. The presence of that content may well mean the lower and middle frequencies were starved for bandwidth and have a lower perceived quality. This is why spectrograms are generally shunned as a means of comparing audio quality. When visitors to the HydrogenAudio forums make these mistakes, they're chided with "we don't listen with our eyes" (well, notwithstanding the McGurk effect).

You might take a look at how I rewrote the caption for your image in the audio compression (data) article. I think it will serve to keep the graph interesting to look at, while avoiding misconceptions among readers.

As for the German Wikipedia articles where you added the same graph, I'm not going to touch them; I try to refrain from editing there, since I don't know the language. —mjb (talk) 01:03, 13 April 2011 (UTC)


 * Wow many informations and... you're right, thanks for the large explication, I will correct it in de:WP and fr:WP as soon as I have the time for it :-) and I think that this -> is also not expressive, what do you mean? Thank you and greetings from germany -- Moehre1992 (talk) 07:15, 13 April 2011 (UTC)