Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: "Sizzling sound" of mp3's is preferred (Read 41340 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

"Sizzling sound" of mp3's is preferred

Some of you may have read it already on Slashdot or elsewhere, but I thought it would be good to have a post here about an interesting article that suggests that more and more people seem to prefer music with encoding artifacts over the original audio or better encodes. Mp3's with noises from bad encodes are apparently so common and widespread that it's much of a norm for how music should sound.

"The Sizzling Sound of Music" by Dale Dougherty - http://radar.oreilly.com/2009/03/the-sizzl...d-of-music.html The article builds upon work and tests by a professor of music at Stanford, Jonathan Berger.

Looking at myself, I can understand how this can happen. I remember having an old cassette with a recording of The Smiths Hatful of Hollow, listening to it a lot, and when later getting a CD of it, it sounded really bad and strange at first because I was so used to the hiss and noise on the cassette. In the article above something like this is also said about people that prefer vinyl over CD, that this preference of crackle and noise from vinyl could be a sort of fetish...


edit:

Here's a link to another post - http://i.gizmodo.com/5166649/ipods-and-you...destroyed-music - where a little more is mentioned how the test was made. http://slashdot.org/article.pl?sid=09/03/11/153205 is the post that got my attention in the first place.

From the gizmodo post, this appears to be a quote from Berger:
Quote
Students were asked to judge the quality of a variety of compression methods randomly mixed with uncompressed 44.1 KHz audio. The music examples included both orchestral, jazz and rock music. When I first did this I was expecting to hear preferences for uncompressed audio and expecting to see MP3 (at 128, 160 and 192 bit rates) well below other methods (including a proprietary wavelet-based approach and AAC). To my surprise, in the rock examples the MP3 at 128 was preferred. I repeated the experiment over 6 years and found the preference for MP3 - particularly in music with high energy (cymbal crashes, brass hits, etc) rising over time.

"Sizzling sound" of mp3's is preferred

Reply #1
I wonder wether those test were blind (let alone double blind) or wether the professor was standing behind his mixing console archly cheering about anyone preferring track B.

"Sizzling sound" of mp3's is preferred

Reply #2
The article is mostly hearsay, being a Professor doesn't mean you do a test the scientific way.

TOS 8 applies

"Sizzling sound" of mp3's is preferred

Reply #3
It actually wouldn't surprise me if people preferred what they were used to. A lot of the arguments people -- who tended to be older -- made in support of vinyl over  compact discs wouldn't have passed HA's TOS requirements but they were vociferous in their partisanship. Why? Because that's what they were used to.

The youngest music listening generation is probably consuming it via lossy MP3 and would be surprised to hear what a properly mastered CD sounds like. I remember as a kid being disappointed by music bought on tape after I heard a song on the radio -- I always seemed to prefer the radio version despite the fact that it likely was compressed and on AM.

 

"Sizzling sound" of mp3's is preferred

Reply #4
Does "the youth" actually listen to poorly encoded MP3s? It's really hard to screw up iTunes or WMP to the point of giving you an encode with the magnitude of badness alluded to in that article. I mean, we just don't live in a Xing 128k CBR world anymore...

My guess is that these kids simply are more comfortable listening to something if they know it is coming from an MP3.

"Sizzling sound" of mp3's is preferred

Reply #5
The article is mostly hearsay, being a Professor doesn't mean you do a test the scientific way.

TOS 8 applies

You said it. As reported, this is totally devoid of any kind of content, except the "mp3 is bad" rant. Maybe the tests were controlled, and maybe they knew what bitrate the lossy files were, and maybe they were checking that the students were really able to tell them apart, but there's nothing here to tell us.

An alternative explanation: he's been using low bit-rate mp3s for a long time. At first, differences between mp3 and lossless were clearly audible, and Stanford music students preferred the better reproduction. MP3 has got a lot better for a given bit-rate, he's using freshly made rips, it's much harder to tell the difference (even for young people with good high-frequency hearing), so "preferences" are getting much more randomized, which shows up as an increasing preference for mp3. From the blog-post, there doesn't seem to be any way to choose between these interpretations.

You don't get to be a professor at Stanford without being real smart, but that doesn't mean you have to be critical of your preconceptions.

"Sizzling sound" of mp3's is preferred

Reply #6
Does "the youth" actually listen to poorly encoded MP3s? It's really hard to screw up iTunes or WMP to the point of giving you an encode with the magnitude of badness alluded to in that article. I mean, we just don't live in a Xing 128k CBR world anymore...


We may not live in that Xing world anymore, thank heavens, but your argument carries the unspoken assumption that everyone rips their own CDs or buys them from iTMS or Amazon. Plenty of garbage still floating around on P2P sites which I would wager still delivers more lossy music files than all the legitimate sites put together.

But I`ll grant you, it`s likely the average quality of MP3s has risen since Ye Olden Days so perhaps I`m labouring under my own assumptions.

"Sizzling sound" of mp3's is preferred

Reply #7
You don't get to be a professor at Stanford without being real smart, but that doesn't mean you have to be critical of your preconceptions.
I was going to issue some snide remark about the fact that he's a professor of music - meaning you really shouldn't trust his opinion on psychoacoustics any more than any other random artist. But looking at his publications, this dude is the genuine article.

That said, there has been some independent confirmation of what was said at the panel. Somebody really ought to email Berger and clarify things.

"Sizzling sound" of mp3's is preferred

Reply #8
I was going to issue some snide remark about the fact that he's a professor of music - meaning you really shouldn't trust his opinion on psychoacoustics any more than any other random artist. But looking at his publications, this dude is the genuine article.

That said, there has been some independent confirmation of what was said at the panel. Somebody really ought to email Berger and clarify things.


I thought that might occur to someone, and I was going to add that of all the humanities subjects, music is the one in which there is least place for fluff--or at any rate, nothing but fluff (former teacher of medieval lit here).

Those articles: hmmm.

He's not first author on most of them--? has sharp grad students? There's a lot about visual presentation of music, including the human genome as music: that sounds like conceptual art to me, rather than psychoacoustics. OTOH, there's a lot about timbre, which suggests that he might, perhaps, have statistical-outlier hearing, and is astonished--perhaps genuinely--that lossy encoding at moderate bit-rates has got good enough to be transparent to a large proportion of the population. One problem is that the articles on which he is lead author, and which are purportedly linked to, come up 404.

We are astonished if people really couldn't tell a live singer from an Edison cylinder, but we are not astonished if, in less than ideal listening conditions, a lot of people can't tell MP3 from a CD. This seems to be, from the blog post axon referenced, what is happening.

It would be nice to have a fuller account.

"Sizzling sound" of mp3's is preferred

Reply #9
I can absolutely relate to this.  I am now a senior citizen.  I can remember back in the early eighties after years of listening to music mostly on cassettes going to a live concert, it just sounded 'wrong.'  I wanted to reach out and turn the treble counter-clockwise.  Logically I knew that the live orchestra didn't sound 'wrong' my ears were simply accustomed to a 'different' sound.  I now try to attend concerts as often as possible.

"Sizzling sound" of mp3's is preferred

Reply #10
I just edited my first post with links to the slashdot post and a blog with what seems to be a quote from Berger, giving a little more information about how the tests were made.

"Sizzling sound" of mp3's is preferred

Reply #11
Quote
I repeated the experiment over 6 years and found the preference for MP3 - particularly in music with high energy (cymbal crashes, brass hits, etc) rising over time.


Cymbals, eh.

Correct me if I'm wrong, but what happens to cymbals in low bitrate MP3 could be considered a form of peak limiting, no? The energy content is smeared, so that the instantaneous peak power is reduced substantially.

Could a valid (hypothetical) interpretation of this result be that the listening volume is so high that the uncompressed samples have peaks so high as to cause an unpleasant sensation, while the MP3 has the most smearing and therefore the "least unpleasant" result? I would imagine that a professor such as he would be using some absolutely primo, uncompressed percussion recordings. That could easily be uncomfortable to listen to at normal college-freshman listening volumes.

"Sizzling sound" of mp3's is preferred

Reply #12
Quote
Students were asked to judge the quality of a variety of compression methods randomly mixed with uncompressed 44.1 KHz audio. The music examples included both orchestral, jazz and rock music. When I first did this I was expecting to hear preferences for uncompressed audio and expecting to see MP3 (at 128, 160 and 192 bit rates) well below other methods (including a proprietary wavelet-based approach and AAC). To my surprise, in the rock examples the MP3 at 128 was preferred. I repeated the experiment over 6 years and found the preference for MP3 - particularly in music with high energy (cymbal crashes, brass hits, etc) rising over time.


I just read this article on Slashdot and my reaction to him is WTF is he talking about? Were they double-blind stimulus tests or pseudo-scientific ones that he pulled out of his ass? I am with Garf the T.O.S 8 approach applies here.


Quote
Cymbals, eh.

Correct me if I'm wrong, but what happens to cymbals in low bitrate MP3 could be considered a form of peak limiting, no? The energy content is smeared, so that the instantaneous peak power is reduced substantially.


I am not an engineer, but isn't it called the Gibbs Phenomenon or is that something different?
budding I.T professional

"Sizzling sound" of mp3's is preferred

Reply #13
BTW, what are these "sizzle sounds" of which they speak? I can't hear MP3 artifacts, and I aim to keep it that way (ignorance is, if not bliss, at least listening undisturbed to the music), but what I've read about sounds more like an absence of sizzle.

"Sizzling sound" of mp3's is preferred

Reply #14
Quote
Students were asked to judge the quality of a variety of compression methods randomly mixed with uncompressed 44.1 KHz audio. The music examples included both orchestral, jazz and rock music. When I first did this I was expecting to hear preferences for uncompressed audio and expecting to see MP3 (at 128, 160 and 192 bit rates) well below other methods (including a proprietary wavelet-based approach and AAC). To my surprise, in the rock examples the MP3 at 128 was preferred. I repeated the experiment over 6 years and found the preference for MP3 - particularly in music with high energy (cymbal crashes, brass hits, etc) rising over time.


I just read this article on Slashdot and my reaction to him is WTF is he talking about? Were they double-blind stimulus tests or pseudo-scientific ones that he pulled out of his ass? I am with Garf the T.O.S 8 approach applies here.


Quote
Cymbals, eh.

Correct me if I'm wrong, but what happens to cymbals in low bitrate MP3 could be considered a form of peak limiting, no? The energy content is smeared, so that the instantaneous peak power is reduced substantially.


I am not an engineer, but isn't it called the Gibbs Phenomenon or is that something different?


That could start an argument, but infact what happens is that the broadband cymbal signal has bits and peices missing, and it gets a metalic, tonal structure as  a result.
-----
J. D. (jj) Johnston

"Sizzling sound" of mp3's is preferred

Reply #15
I'm not surprise to see some people preferred the "smoothed" version of some rock music. I don't like it but I can see why some peoples like it.

If peoples start to like the sound of artifact in jazz, orchestra, ETC.. Now that would be really interesting.

Come to think of it, I really want to know what the encoder use to created 128 kbps rock tracks. Last time I tested LAME 128 it was pretty good and didn't have "Sizzling sound" that much. May be they use BLADE or something.. 

"Sizzling sound" of mp3's is preferred

Reply #16
It was this "sizzling sound" on other people's CBR128 MP3 encodings that put me off using MP3 for many years. It wasn't until I looked into how to make MP3 encodings properly myself that my mind was changed. Maybe the Prof should be making it clearer that it's abuse of the standard that can lead to clearly audible artifacts rather than the standard itself?

Cheers, Slipstreem. 

"Sizzling sound" of mp3's is preferred

Reply #17
So there are two possibilities.

1. He's using old, bad MP3 encodes which are definitely NOT transparent, but for some reason undergrads have got more tolerant of this effect. This is what he's claiming, but the prima facie problem is that while his examples might be crap, it's unlikely that this year's freshmen have heard such bad files; as the recent ~128 kbps listening test has shown, the most popular MP3 codecs have all more or less caught up with LAME, and the other really widespread format, AAC, is at least no worse. So how could the studes have got used to a lower standard?

2. When he first did the tests and listened for himself, there really was a marked difference, which both he and his students could hear. As the years have passed, he's used recent encodings, which have got progressively better, but he hasn't noticed, for a variety of reasons which could affect all of us. His students, however, in many cases can't tell the difference, hence the rise in preference for MP3 (if the test is properly blind, one would expect people who can't hear a difference, but are constrained to state a preference, to choose the MP3 about half the time. If the MP3 is transparent for half the group, then you'd expect about a 25% preference for the MP3).

"Sizzling sound" of mp3's is preferred

Reply #18
But bear in mind that the recent ~128Kbps listening test was using VBR, not CBR, so a bitrate of up to 320Kbps will have been available to most (if not all) of the encoders as and when deemed necessary by the encoders on-the-fly. It sounds to me as though the Prof is deliberately strangling whichever encoder he's using by encoding in CBR.

CBR128 encodings even with the current version of LAME still exhibit the "sizzling sound" and transient crushing to my ears with some specific musical content (eg, rock with heavily 'ridden' cymbals), whereas a VBR encoding averaging 128Kbps doesn't, or at least not so badly. FYI: The "sizzling sound" equates to someone rubbing my ears out with sandpaper whilst I'm listening and used to give me migraines after short periods of listening, although I can tolerate CBR128 a decade on if there's no alternative. LAME in VBR at -V3 (~175Kbps) always provides me with stress-free listening even if the results aren't always entirely perceptually transparent.

I agree with your second point entirely and believe that the improvement in encoding quality even in CBR at sub-optimal bitrates may explain why the students have become more accepting over the years. I don't think the Prof is proving anything that most of us here don't know already, but I do disagree with the way he's interpreting and presenting the results.

Cheers, Slipstreem. 

"Sizzling sound" of mp3's is preferred

Reply #19
There's an entire generation growing up thinking mp3s are what music sounds like.  Depressing really.

"Sizzling sound" of mp3's is preferred

Reply #20
There's an entire generation growing up thinking mp3s are what music sounds like.  Depressing really.


Oh well, I grew up thinking that AM radio and low quality cassette tapes were what music sounded like, so I guess they're the lucky ones. Especially if they're using good quality encoders at a decent bit rate.

"Sizzling sound" of mp3's is preferred

Reply #21
There's an entire generation growing up thinking mp3s are what music sounds like.  Depressing really.


By that do you mean unidentifiable from the source material?

"Sizzling sound" of mp3's is preferred

Reply #22
There's an entire generation growing up thinking mp3s are what music sounds like.  Depressing really.


By that do you mean unidentifiable from the source material?


No. The kids subconsciously preferred the lesser quality file. Most people don't use good encoders like the people here at HA. As well, many listen on terrible quality earbuds. I should have been more precise. People used to obsess over the best possible stereo equipment they could find. Now the emphasis is on not having to be bothered with listening to the music, but rather how convenient is it.

"Sizzling sound" of mp3's is preferred

Reply #23
I think the kids now with moderate grade earphones are the equivalents of the kids of my generation, who grew up on 45 rpm singles played with a ceramic cartridge through one dodgy speaker. And some of them have grown up to be the farts who buy Sun Studio Presley recordings, or the 1951 Furtwaengler Beethoven 9, or whatever. Seriously, I think top-quality reproduction has never been high on teenage priorities, for as long as there have been teenagers, and it's not always the top concern for music lovers.

Enjoy.

"Sizzling sound" of mp3's is preferred

Reply #24
There's an entire generation growing up thinking mp3s are what music sounds like.  Depressing really.


By that do you mean unidentifiable from the source material?


No. The kids subconsciously preferred the lesser quality file. Most people don't use good encoders like the people here at HA. As well, many listen on terrible quality earbuds. I should have been more precise. People used to obsess over the best possible stereo equipment they could find.



Only some people.  And only some people still do.


Quote
Now the emphasis is on not having to be bothered with listening to the music, but rather how convenient is it.


apparently you've forgotten

--portable radio
--the 45 single and 'portable' devices for playing them
--the LP/45 'changer' where you stacked discs up
--car stereo
--the cassette
--the walkman

and that's all before the CD revolution