The Ghost in the MP3 (2014) | Hacker News Reader

The Ghost in the MP3 (2014)

theghostinthemp3.com

35 pointsTomte16 comments

> primarily designed by and for western-european men, and using the music they liked

What is the point of this sentence? Any value this may have as a technical or art piece is lost by the random attack on the creators and their race.

I'm not even convinced that it is true. I was under the impression that they chose music specifically that was difficult to compress well.

It seemed like they had a larger social justice angle for the piece, then realized it didn’t read well or forgot to write it in the first piece.

To be honest, I expected the article to go that route just from that statement, but refreshingly it didn’t. While valid, the social justice argument is hard to take seriously - audio engineers are people, too, and you can’t expect everyone to be aware of every single one of their biases. Especially not back in the times when MP3 was created.

Still, it would be interesting to read about the technical differences between music in different cultures. As technology spreads in the developing world, accommodating all cultures will grow as an issue.

As MP3's have invaded more and more contemporary listening spaces, the class of privileged sounds which the format inadvertently creates has become more apparent... This format has become a curator for these spaces: allowing in a great deal of wonderful sound, yes, but at the exclusion of a vast territory in the available sonic terrain.

Wondering what kind of creative expressions have been blocked by the status quo is absolutely social justice. Just admitting that MP3 might be holding back entire cultures or subcultures of music is going to be a big leap for some people.

So what wording should people use when they want to suggest that someone's culture constrained their thinking, without triggering the anti-SJW brigade? Some of those people are so easy to accidentally trigger and so sensitive. How could we have a conversation about this without upsetting them and the conversation subsequently degrading?

Instead of originating with gender and trying to fit in taste of music etc. into that. Start with what is relevant, the music, maybe that happens to fit nicely into what is considered popular for a specific group of people. Maybe not.

"It seems they targeted this kind of music which would have been quite typical for XXX during that time."

Or mentioning more specifically that they omitted this kind of music which could explain why it perform poorly in that context.

The quote also reads that they it was designed explicitly for western-european men, the gap between that and "they might have biases toward their own taste" is astronomical.

Sounds like a descriptive non-combative sentence to me. Why do you perceive it as an attack?

The whole tone of the piece has a kind of "fight the man!" vibe, talking about putting up an "act of resistance" because "MP3's have invaded more and more contemporary listening spaces".

You've got an institutionally-accepted construct designed by white European men for their uses, which inadvertently creates a privileged class that implicitly excludes anything that would've been "other" to those designers.

I wouldn't call it an attack, but I'd call it a weird and not well supported (by what's posted here) social justice angle.

The point is context. The technology wasn't created in a void. I wouldn't accuse the sentence of being well written, and the point could have been made more clearly. Simply saying that the developers used psychoacoustics were tested against the developer's own hearing preferences and tested against their music preferences. A wide range of different listeners were not consulted and a wide range of musical styles were not tested during the development phase. All technology has a context. Pointing out context isn't an attack.

I agree with your point but how is "western-european men" an appropriate context? What does that imply? Isn't reducing the taste of the developers to their gender sexist?

> A wide range of different listeners were not consulted and a wide range of musical styles were not tested during the development phase

Luckily, codec developers aren't totally stupid. File formats tend to be defined through their reading operation, not the writing operation. Fraunhofer MP3, LAME and many others went way beyond the initial mp3 reference codec in terms of deciding what parts of the signal are important or not.

Synthetic tests like the "put noise in it, check if it's still evenly distributed in a frequency plot" one the article shows are misleading. I could tweak lame into recreating regular noise sufficiently, while still creating valid MP3 files. The same input probably looks quite different when put through 8hz-mp3 or some early bladeenc (which are similar to the ISO reference codec, and _very_ unlike contemporary LAME's psychoacoustic model).

Optimizing your record studio output to any given codec version is futile for that reason, too: Do you optimize for 1998's MP3 encoding? or 2015's? or Vorbis, AAC (and which of its 2 dozen profiles?), Opus, ...?

Doubt it was meant as an attack more a statement of fact. If I remember correctly all the songs used to benchmark the compression were of "western-european" origin. Same thing for the people judging the result.

Here is an interview with the creator, mostly about the ghost in the mp3 project http://traffic.libsyn.com/dataskeptic/The_Ghost_in_the_MP3.m...

Once again a "null test" like this, completely disregards the fact that lossy codecs such as MP3 use a psycho-acoustic model to determine which frequencies can be discarded with minimal impact to what humans actually hear, and is therefore misleading if considered as a means for judging the audio quality of a codec... mildly interesting as an art project though.

I think that is literally what the project is about.

Just posted for clarity - I remember people getting the wrong end of the stick when it first became popular.

Previously at https://news.ycombinator.com/item?id=7955917, https://news.ycombinator.com/item?id=9049196.

This project seems to only use the LAME encoder, but it would be interesting to see if other codecs give different creative results. And now that the patents have expired, someone could even write their own codecs from scratch with interesting artifacts in mind.

Reply on news.ycombinator.com