The effects of musical genre upon ease/degree of compression, [split from “What music genres do you listen to?”, thread 96042] Options

[Porcus]

Can we use that to excuse errors in subject-verb agreement?
http://www.hydrogenaudio.org/forums/index....&pid=802685

If so I'll see if there's room to amend my sig.

I can has pluraliz majestatiz?

[dhromed]

And he said that he could "metalicize" the harpsichord sample and make it even harder to compress. I showed that adding noise or distortion etc actually makes it easier to compress (when mesaured with bitrate at the same q setting).

Have you tried the same harpsichord treatment on FLAC? With lossless compression, adding noise and distortion should increase file size compared to the untreated compressed version.

My point is that "easy to compress" with lossy compression shouldn't just mean a bitrate reduction. I could zero all the samples. Or I could delete the file. Now that's compression! Of course there's zero quality left at all— but the compression was really easy!

In other words, with lossy encoding, something should only be called "easy to compress" if you can reduce the bitrate without introducing audible distortion. If you've manually added noise and distortion to the harpsichord, clearly that is not the case. You've only made it easier to compress by making the lossy-encoder's work easier, thus supporting your friend's point that metal is hard to compress (regardless of whether this is correct, mind you. Experimental results can support a false conclusion).

I hate to raise a comparison to GIF and Jpeg, but in the case I think it's merited: Photoshop's Save For Web has an option to add special noise to an image that makes it easier for it to be compressed as GIF. It often reduces quality but can also significantly reduce file size. Similarly, you can add a tiny bit of blur to an image destined for jpeg, and thus reduce filesize.

It's a form of tenderizing the meat, so to speak, before it's fed into the compression algorithm.

This post has been edited by dhromed: Jul 20 2012, 09:45

[Porcus]

Well I just got stunned by a sample that certainly isn't harpsichord, but which Lame VBR thinks is way worse: http://www.hydrogenaudio.org/forums/index....st&p=802867

[greynol]

If adding a small amount of noise/distortion/error/whatever to harpsichord made it easier to compress but didn't change the sound then that would have been interesting. If it does change the sound then it is not; nor is the conjecture about why metal is hard to compress, which I don't believe to be true (that is to say the conjecture, not that metal is difficult to compress).

Personally I'm a bit bothered about seeing more comparison (EDIT: of lossy encodings) based on bitrates alone without double-blind testing these days. For situations where samples sound different there is still ABC/HR (in case you want to discuss them objective TOS8-compliant way).

This post has been edited by greynol: Jul 20 2012, 18:12

[greynol]

With lossless compression, adding noise and distortion should increase file size compared to the untreated compressed version.

What if the noise is correlated?

This post has been edited by greynol: Jul 20 2012, 15:51

[pawelq]

Again, the purpose of that test was to address a specific assertion. Which was that rock/metal is harder to lossy-compress than classical because it contains more distortion, noise, non-harmonic and/or mistuned components. If this was true, then applying distortion/noise/mistuning etc to a harpsichord sample should make it more difficult to compress than the original harspichord sample. As much as the resulting bitrate at the same ogg quality setting is a measure of being easy vs. hard to compress, by small test showed the opposite. The modified sample was not supposed to be audibly indistinguishable form the original non-modified version, same as adding heavy distortion to a guitar is supposed to be audible.

This lead me to a conclusion that other factors than the amount of distortion and inharmonicty may be influence lossy compression.

[Porcus]

Again, the purpose of that test was to address a specific assertion. Which was that rock/metal is harder to lossy-compress than classical because [...]

Oh?

I had a similar conversation before with people who claimed that heavy rock/heavy metal would be more difficult to compress than "classical" due to distortion, inharmonic content, etc. They claimed they could "heavy-metallize" my harpsichord sample by adding more distortion/noise/mistuning (the argument was in heavy metal you have two guitarists who play slightly differently) and that sample would require even higher bitrate. Thus supposedly "proving" that harpsichord is not as hard to compress as distorted heavy metal.

[emphasis mine]


There is no refefence here to lossy, not before you started referring to a «test» that -- it seems from your posting -- was made up by yourself.

[pawelq]

There is no refefence here to lossy

Sorry if I caused a confusion because of my skipping the word "lossy" there. I however said just below "I was able to show that adding distortion and overlaying another slightly mistuned and time-shifted copy of the same harpsichord piece actually reduced bitrate for the same quality of Ogg Vorbis (160 kbps vs 195 kbps). "

I would post a linkto that discussion to make sure that I don't inadvertently skip any relevant information while reporting what has been done and said there, but it wasn't in English.