Hacker News new | comments | show | ask | jobs | submit login
“MP3 is dead” missed the real, much better story (marco.org)
650 points by imartin2k 42 days ago | hide | past | web | 290 comments | favorite

Wait, people thought mp3 was dead because of that announcement?! O_o I feel like people have lost touch with the history and the entire reason many Linux distros do not distribute mp3 codecs!

Some of these writers/editors really need to spend each day reading at least one headline from that day in 2016, 2015 .. all the way back to 1999.

MP3 is now more free/libre, or at least when it comes to encoding/decoding. Patents are one of the reasons people have been so hesitant about H.264 and why we see things like WebM.

Personally, I've encoded in FLAC for years, and even in the early days (2003-ish) I was backing up CDs to oggs instead of mp3s. Unless you're really concerned about space today, download your music in a lossless format. Sites like Bandcamp and CDBaby now support lossless (FLAC and ALAC)

People (some/many) believe what they read. Some CEOs in some corners of the planet will rush to their CIO and tell them "I don't know what this mp3 is but lets switch to aac A-S-A-P!!!". Then the poor CTO (if he/she has a technical background) will have to take 15mins to explain to the CEO that "mp3 is not dead, the patent is gone so not it's absolutely free, while for aac we'll have to be paying ABC amount for our XYZ product/service".

I guess the current aac patent holders are smiling now :)

Confuse & conquer!!!

The real shame, IMO, is not that some CEO or other non technical person is confused, but that reputable news outlets reported it that way. (For some value of reputable).

If you look at the three links that Marco posted in the first sentence, they lead to articles by Gizmodo, Engadget and NPR, all of which completely got the story wrong. That's shocking, if you ask me.

Gizmodo and Engadget are not "reputable news outlets", but junk blogs for users of browser toolbars with announcements for each chinese tablet. I never seen any half-decent article on such sites. It's even stupidier than clickbait, there are no sensationalism and attention manipulation, these sites are just content fillers, their authors just compose words randomly.


It's not just Engadget and ilk. The difference between being end-of-life as a product for Fraunhofer, and obsolete as a format, is a distinction not many outlets have made.

Is there even a reputable techinical news outlet? TechCrunch is too very clickbaity and editorilized, although gets linked here all the time

Ars Technica, perhaps?

Despite other comments here, I'm as happy as ever with Ars, and I've been reading it for many years. They're the only tech news site I read anymore (besides here of course).

I've generally seen people get on the wagon of bashing Ars then not having anything better as a replacement suggestion. Like yeah I know the actual blog source is better but I don't know every tech blog on the internet.

No. Absolutely not. After they were acquired by Conde Nast, it has been downhill at an amazing pace.

I have always found Ars to have relevant and well written content. Conde Nast has a lot of good writers at their various properties.

They were downhill well before Conde Nast. Almost every article they ever did on LEDs was horribly out of date, using information that was true maybe a decade ago.

Eh, it's been a long time since I read it. Maybe it was for the better.

I used to frequent Anandtech, Digital Foundry and ExtremeTech as they all had pretty good articles that sometimes went quite deep into the matter at hand. Anandtech its quality has gone down a lot, ExtremeTech is full of ads, ZergNet links and a constantly running JS script (probably a tracker?), and Digital Foundry its deep dives have become a lot less frequent. Only Ars Technica remains :/

Anandtech is still very good imo. I haven't really found any website that compares.


It's an aggregator, and you need to resist the urge to click on the links. But the headlines and blurb are often informative enough that you usually don't need to click.

> you need to resist the urge to click on the links


Techmeme heading/summary:

"Fraunhofer, the major contributor to MP3, shut down its licensing program in April as MP3 patents expired — Red Hat has announced that Fedora will include official MP3 decoding and encoding. The reason is that MP3 is now patent free - as far as anyone can tell."

First linked (Engadget) article: "MP3 is dead, long live AAC. Its creators have abandoned licenses to the format, signing its death sentence."

Not clicking would have both saved you time and left you better informed.

Well they are not famous for their news articles but their technical reviews. If you are looking for a laptop or a phone, then gizmodo and engadgets are really good sources. My guess is that they had to expend (investors probably asked for it), so they did news (badly). But their core skill is still very good.

I stopped reading Engadget the moment I saw this tag used for the Nikon D810A DSLR for astrophotography in 2015 [1]: "Astrology". <facepalm/>

[1] https://www.engadget.com/2015/02/09/nikon-d810a-astrophotogr...

Clearly a typo, probably due to autocompletion in the tag selector widget since : https://www.engadget.com/tag/Astrology/

Convincing a reasonable CEO that news sources can be garbage should not be difficult.

>the poor CTO (if he/she has a technical background)

A CTO without technical background is indeed in a very poor state

The aac patent holders are the same who wrote that mp3 announcement — mpst of the patents on aac are held by Fraunhofer again ;)

Fraunhofer is, conveniently, one of the AAC patient holders.

big companies have "we need licences for everything" people.

There was one on Australian national TV this morning saying we will have to start using something more modern like mpeg.

mp3's are the jpeg of the audio world. Maximum compatibility with any and all ancient electronics, and now totally free. I don't think we'll ever be rid of it. And really that's fine because at high bitrate mp3 is very usable and mp3 does a great job at what it does like jpeg does: it dramatically reduces the data size compared to raw (wav to mp3 savings is dramatically greater than mp3 to any other audio format savings).

Where "very usable" == "sounds exactly the same as lossless", unless there's some double blind study showing that people can reliably tell the difference between v0/320 mp3's and lossless... I've never seen one.

As a distribution format for mixed-down music that will be consumed from a portable device's internal storage, 320kb/s VBR is great.

There's some unfortunate irony in that we just got out from under the MP3 patents, which mean that MP3 encoders can finally get mainlined and won't require screwing around like we're all gotten used to for a decade+ (having to add some extra repo to your apt.sources or equivalent file, etc.), but at the same time the consumption patterns of music have shifted again, away from portable devices with high-density storage like iPods, and towards streaming particularly over cell-data networks.

There is an argument, although I do not like some of its conclusions, that some of the newer codecs are preferable to MP3 if you are working within <96kb/s for a stereo signal so that you can put it over a 3G connection or something. MP3 starts to sound audibly bad at lower bitrates, depending on what you're listening to.

My cellphone today is a much more powerful device than my 2007 iPod, but doesn't have nearly the same amount of available onboard storage (and what storage it does have is split between music and photos and software and other stuff). So the tradeoffs that make for an ideal codec choice on the iPod aren't clearly great on the phone.

IMO, this is an argument for keeping your personal music collection in an uncompressed/losslessly-compressed format, so that you can recompress it easily based on how it's actually going to be used.

The bigger the system is, the noticeable the difference is. It's the same as displaying jpgs versus analog images on a large projector. The difference is much more noticeable with the audio! But yeah, this is a more of an analog versus digital debate, rather than comparing digital formats. However lossless is called lossless for a reason.

It's more about branding. In the early days, an mp3 could easily be some 64k cbr shite.

If you were sharing, FLAC was guaranteed high-quality and free. AAC/WMA usually meant copy-protected though likely a higher quality.

>In the early days, an mp3 could easily be some 64k cbr shite.

When I was a teenager my first MP3 player (RCA Lyra) came with one 32 MB CompactFlash card. I cut the bitrate down far enough to fit 4 or 5 full albums on that at once. It sounded like absolute garbage but I was the only kid who could fit that much music in my pocket.

I tell this story because I'm still impressed by what MP3s can do after so much time.

One downside to MP3s is that it's not possible (or maybe not always possible, or not easily possible) to prevent silence from being added to the beginning and/or end of the audio. This complicates things like gapless playback or loops. It also complicates using audio where a high level of precision is required (e.g. "I want this widget to appear exactly half a second into this audio").

Gapless playback is a thing with MP3s, but it seems to involve specific support for the encoder. I find this quite sad.

Ogg Vorbis is superior in every way, and I really don't care about ancient electronics, so I'm perfectly happy encoding all my music into .ogg since all my devices, including my car, are compatible with it.

"Ogg Vorbis is superior in every way [except ways] I don't really care about"

The only way it's not superior is compatibility with ancient electronics, just like I said before. What's the problem? Why should I care about playing my music on 15-year-old hardware? I don't have any such hardware any more, so I really don't care about it. In fact, 10+ years ago when I still had a portable music player, I had an iRiver H340. Even it played my Oggs!! So where are these shitty players that can't play Oggs? I've never had one. And these days, they don't seem to exist. My Android phone plays them just fine, out of the box. My car even plays them right off a USB thumb drive. And of course my computer plays them easily. So again, why should I give two shits about compatibility with some ancient devices that I don't have, and likely not many other people do either?

Do you only use software that'll run on Windows 2000?

> Do you only use software that'll run on Windows 2000?

Mainstream support for Windows 2000 ended twelve years ago.

> So where are these shitty players that can't play Oggs? I've never had one.

Apple is selling a music player called "iPod" right now. Perhaps you've heard of it. Updated in 2015 https://en.wikipedia.org/wiki/IPod_Touch_(6th_generation)

Guess what iPods don't support? OGG.

Guess what iPods do support? MP3.

If you want to buy some shitty portable music player that can't handle proper open-source file formats, that's your choice. I've had a bunch of devices from all sorts of other, better manufacturers that ALL supported Oggs.

BTW, "Ogg" isn't an acronym.

If you call "number of people using the format"part of the featureset, mp3 is superior in one aspect at least.

Network effect

Similarly, Loglan is superior to English in every way.

If all English speakers also spoke Loglan, except for the tiny portion of the population that's over 100 years old, then your analogy would make sense.

honestly thought you were being sarcastic when i read that

I still use mp3s - compatibility is generally better than any other format and take up a more reasonable amount of space than lossless formats. I really don't hear a difference between an mp3 and a lossless version, and that is with a much better ear than most non-musically trained (I have given useful critiques to artists on WIPs related to mixing & mastering). In addition, I have over 1 TB of music that is almost wholly mp3s - it's not terribly practical for me to have that in lossless either.

I've heard some explanations on why lossless is different even tho the stuff cut out of high quality MP3/ogg is in frequency we don't hear, but it honestly sounds like "high end cables/pure gold contacts" to me.

At high bitrates maybe mp3 has the same heard quality. But mp3 has his artifacts, and the problem is if you heard them once, you couldn't stop youself from hearing them every time. It's like interlacing in movies or bugs at borders of contrasting areas of picture in jpeg. And yes, this can be proved by double-blind experiment at least for 128 kbit/s mp3. With 192 kbit/s, I believe, you can hear artifacts also, but you would need some special training for it. So with 192 kbit/s solution is simple: just do not teach youself to hear artifacts. With lower bitrates it's not so simple.

As for me, flac is better for storing sounds. Stored flac has no artifacts (if obtained without them), I can recode it to mp3, vorbis, opus or something else, and get artifacts from that codec, but not artifacts from mp3 plus artifacts from that codec.

Can you give some sources? What are these artifacts you talk about?

And even assuming that it's audible at 192kbit/s with a decent encoder, why not just use 320kbit/s MP3 which are still significantly smaller than lossless?

16bit lossless audio is great for producers, you have a lot of headroom for messing with it, re-encoding it, twisting it, amplifying it, mixing it etc... But if you're just archiving public releases for "consumption" it's completely overkill.

>It's like interlacing in movies or bugs at borders of contrasting areas of picture in jpeg

In what way? Interlacing is completely irrelevant in the context of audio.

I'm sorry if I come off as adversarial but it's a pet peeve of mine. The maths and physics behind signal processing are very well understood, if some algorithm induces a signal degradation in the audible spectrum it should be trivial to show it objectively with the right measuring equipment and/or the right mathematical equation.

> What are these artifacts you talk about?

Splash cymbals in particular sound terrible (or noticeably degraded) on low bitrate stereo MP3s (ie 128kbit/s). Some early encoders actually did a lowpass filter to eliminate frequencies above 16kHz, to mask the 'watery' high-frequency warble (see [1]). Note how even the Fraunhofer codec differs a lot at high-frequencies in the charts on that link.

Some of it is still noticeable even in lower bitrate AAC. I've got U2's Pop album on CD from 1997 (coincidentally the first album to be leaked as MP3 [2]), and even at 128kbps iTunes AAC the cymbals just don't sound right on 'Gone'. They're watery, like an audio version of a badly compressed JPG. I keep meaning to re-rip it at higher bitrate.

[1] http://archive.arstechnica.com/wankerdesk/1q00/mp3/mp3-3.htm...

[2] http://pitchfork.com/thepitch/652-a-history-of-digital-album...

https://en.wikipedia.org/wiki/Pre-echo is the usual example. I did hear it in that audio example on Wikipedia, but I doubt I'd hear it under realistic circumstances.

If I remember correctly, some of it came down to encoder settings as well. Joint stereo, for example, always sounded janky at all bitrates. CBR vs VBR also made a difference. Also, the quality of the source was a factor. A low quality source that was already kind of lossy would sound much worse than the source itself when encoded at lower bitrates.

This. Maybe people have different hearing skills also. But I can definitely hear the artifacts in MP3s, like hihats getting muffled and the deep base getting lost in the sea of MP3 compression.

With good headphones (I'm talking 200+ euros class headphones for example) there is a certain difference between even a 320 kpbs MP3 vs a 128 AAC.

AAC or OGG for example just has better usualy clarity in the sound, the bass is clearer and more distinct and the hihats and other hi sounds aren't muffled in the background or don't lose their clarity like they do with MP3s.

And like the parent wrote, once you have heard those 'signature losses' of the mp3 codec, it's very hard not to notice them.

320 mp3's sound just like .wav's to me. I have a middle of the road studio at home, and they sound great on my 8" monitors. I'll play back to back albums in wav, then mp3. Nothing sticks out at me.

who uses 128 kbit/s mp3 these days? the lowest I find are 192

I have a ton of CDs I ripped to 128 kbit a long time ago, back when I couldn't afford the space for anything larger. Honestly, at my age, with only a few exceptions, I don't really notice anything.

Still, when I get new stuff, or rip now, I try to do it at least at 192 or 320, since space isn't an issue any longer.

From my perspective, I don't think I'd ever be able to tell the difference between high quality MP3 and FLAC in an ABX test.

If I first heard the MP3 version, and then heard the FLAC... I might be able to point out which one was which afterwards, under very good conditions and depending on the recording. I remember doing this test several years back. The general details are pretty much the same, but the MP3 version kind of changes the "sound stage" a bit for a lack of a better term.

It's very subtle and in practice, you need critical-listening environments even to do this. 99% of music is not heard in critical listening music environments. A huge amount of music isn't even engineered for this kind of critical listening these days (eg: you need recordings that haven't destroyed the dynamic range in the first place, which rules out most radio pop today.)

I agree that having the ability to transcode to lossy with as little loss as possible is IMHO the best reason to store music in FLAC.

I agree with all of this - I have had studies quoted to me by artists about self-proclaimed audiophiles performing worse in a double blind test identifying which was the lossless audio vs. a high bit rate mp3 (256 kbps? I forget). The evidence seemed to suggest quite strongly that humans aren't able to distinguish either.

I would love to have lossless audio purely for storage purposes - however, it would be quite expensive for me to do so with the amount of music I have, as well as time consuming. For me at least, mp3 is the perfect balance, and its ubiquitous compatibility makes it a no brainer.

I have to wonder how many who claim to immediately tell the difference between mp3 and FLAC have instead simply been exposed to horribly encoded mp3s.

Although this is anecdotal, it pings off your own experiences: A friend of mine (and self-described audiophile) recently got into a discussion with a mutual friend regarding mp3, FLAC, and how both of them could tell the format apart simply by listening. I suggested that perhaps the mp3s he had available were simply poorly encoded, and sent him a sample of my library that's a direct rip from a CD (at a modest bitrate of either 192kbps or 256kbps, I can't remember). He then realized that the mp3s he had been comparing against his FLAC copies of the same album were likely YouTube rips from before he bought the album, and that he likely never downloaded the official mp3s when he purchased the FLAC sources. He confided that he had a difficult time determining which was which when comparing the two on more equal footing with the sample I sent.

I still posit that the average person can't tell the difference, even with above average hardware. Exceptionally sensitive hearing outside the normal range of most humans notwithstanding, a decently encoded mp3 (or Opus format for that matter) should hardly be distinguishable from a FLAC, if at all. I suspect that most of those who claim they can hear the difference aren't particularly special--they're just comparing poorly encoded files against a source format.

At any rate, I'm in my mid-thirties, and I know my hearing isn't quite as acute as it was a decade ago. FLACs are more of a novelty for the sake of original sources, because there's no way I could tell the difference between that and mp3s with a good bitrate. With the exception of the Hunt for Red October soundtrack, I've not been able to tell low/moderate (~128kbps) bitrate mp3s apart from high (384kbps) ones AFAIK (the high pitches in the hymn "warble" at lower bitrates--not sure I can detect this now, though).

I don't store lossless because I think I sounds better. The idea for me is I can transcode to whatever format I need without generational quality loss.

I own "high-end cables" (not sure if the contacts are gold) and my low-end-of-high-end HiFi might sound just as good with cheaper cables but I don't know, having never (yet) tried it that way. It's very likely that all that super great sound is just the high quality amp and speakers.

The same thing could be happening with lossless. If you are going to the trouble of encoding and storing a lossless digital music collection, I think chances are good you're listening on very good equipment. And you might scoff at the idea of trying out high-bitrate MP3 on it, because the MP3s you have from your Napster days really do sound terrible through your tube amp.

OTOH I wouldn't go too far down the "can't hear it anyway" road. I'm in my 40's and there's plenty I can't hear, but I assume not everything in my collection will be on YouFacePotify for my grandkids to listen to while their ears still work.

Yep. I'm generally only interested in good recordings and don't particularly care about disk space, so MP3 is perfectly fine. I'm not an "audiophile", but I do keep lossless when possible to stay future-proof.

My scheme is FLAC when I can, otherwise MP3. I can render the FLAC to MP3 (a batch script keeps things up to date) for compatibility/size with some things, but my main system plays FLAC just fine.

It isn't practical to take my full collection with me in MP3. Just checked, and the music share is currently 5.38T, but that's with a full set of MP3 dupes. And there is some garbage in there I haven't has the patience to weed out yet. I just point the phone at the MP3 directory and pick 'n choose with ITunes. If I really want some random song while not at home, I can pull it off my server easily enough, but at my age, I don't really develop strong needs for a particular song.

Every other reasonable person on Earth, and I, agree with you. Lossless if a fine this to have when possible, or for something you might care about. But considering the number of times I've had to incorporate a Thomas the Engine or Weird Al song to my library for my son, well, I have other concerns in life than maximum auditory experience.

I keep my main collection in source format (preferably flac), then transcode it down to mp3 or opus depending on my target (eg. my phone can handle opus but my ancient USB mp3 player can't). In particular, it's important that I can use opus on my phone because I want to fit as many songs as possible, and by my own personal listen tests, I need to double the bitrate for an mp3 to sound as good as an opus (though granted that wasn't blind tested or anything).

I think they were suggesting lossless so that you could convert it into whatever format you need.

Personally I encode into any format I please and to hell with patents. Let em sue me.

Tech blogs use each other as sources, so if one misunderstands something it can go around as fact for some time and show up on many blogs.

People thought mp3 was dead because of they have lost (or never had one) critical thinking and believe everything (or just too much) that clever PR people try to convince them to believe.

I can understand why Fraunhofer made this statement (it is still despicable) but it is really sad to see how gullible people are.

Sites like Bandcamp and CDBaby now support lossless (FLAC and ALAC)

I haven't used it in years, but I'm pretty sure Bandcamp will only accept lossless codecs.

Right, but of course they support downloading in other formats (ogg, etc)

FLAC sucks because no metadata.

Why should it bundle the kitchen sink - I wouldn't expect all codecs to come with their own container format! You can put it in Ogg or use its own more-limited native metadata handling.

> I wouldn't expect all codecs to come with their own container format!

Huh? That seems totally wrong to me.

I wouldn't expect a codec to develop their own, unique, incompatible container format, and I would hope very much that they would not (!!), but it's a real mistake to develop a codec and not specify at least some sort of preferred container format.

Otherwise, people are going to do what is often done with FLAC: not put it into a container format and just treat the raw output from the codec as a distribution format. And then complain that the whole thing sucks because there's no support for multiple streams, metadata, lyrics, subtitles, album covers, cuesheets, whatever.

The way to avoid that is not to set users up for failure by releasing the bare codec without some sort of preferred container format that's used by default. If the FLAC encoder had always used OGG containers (or Matroska or Quicktime or AVI or whatever) by default there'd be a whole lot fewer bare FLAC files floating around, convincing everyone that "FLAC sucks because there's no metadata".

That said, there are some issues with container formats as they are frequently implemented, which tends to drive users towards not using them and just exchanging bare codec output: if you separate the user from the codec with an intermediate container layer, you can create a ton of frustration when users get files that they think they can use, based on the file extension, but then get an error because they don't have the codec du jour... but there was no obvious way to tell that when they were looking at the file. The codec used is far more important to the average user than the type of container format, most of the time. (And yeah the ideal solution would be for filesystems to stop sucking so badly at metadata; MacOS and HFS was better at this stuff in 1996 than most modern computers are today -- at least it had the idea of both a file "type" and its "creator" as distinct things from the extension.) But in our world, the result is some containers being perceived as "unreliable" or "fussy" because they're used for a diversity of codecs that not all implementations have.

I don't have a great solution to that second problem, but at the very least I think that the file extension should follow the codec combination used inside the file, and not the container format. E.g. an AVI container with AVC compressed video inside shouldn't have the same file extension as an AVI container with Sorenson Video inside it; those two things are not interchangeable as far as a user is concerned. Since file extensions are the only metadata users get, they need to somehow represent the combination of both codec+container.

MP3 is really, really​ smart. Like seriously.

Not sure why anyone would regard a patent expiration as a sign of its death. It should be the opposite! It's now free for everyone and in the public domain, and that is cause for celebration.

It's a bit surreal. Engadget interprets "becoming patent-free" as "being retired". Gizmodo interprets it as "it's dead". We're not in a world where file format fanboys have the writing prowess to hold positions in the media right?

Look for companies with financial interest in locking down media and playback and patent cartels that want you to pay for their codecs and you will find your "fanboys". MP3 is now both license and drm free, anyone can write it and play it anytime and anywhere they want without paying a cent. For some people that thought is nightmare inducing.

> For some people that thought is nightmare inducing.

Rhetorical questions...

Who are these people for whom the thought of not making that money causes them such stress and anguish?

...furthermore, what were they like as children?


This seems like a natural if pathological outgrowth of tech culture obsessed with newness. When a 3 year old phone is described as ancient, obsolete, or out of support for two years, how good could a 2 decade old compression system be?

It's so ancient that it makes you assume that it has to be truly exceptional to survive that long in such an environment.

Exceptional might be the wrong term but I think there's a plateau effect: MP3 was the first mainstream codec to reach the level where most people didn't perceive any drawbacks and there's a huge inertial benefit.

I don't regret ripping my entire collection as AAC / lossless but I also acknowledge that the space savings are basically a waste of time for anyone not using good speakers / headphones in a very quiet environment, and outside of a handful of pieces susceptible to pre-echo lossless was a hedge against double-encoding in the somewhat unlikely event that the future codec landscape changes dramatically.

Modern compression schemes are better, but for most people the difference is academic.

No argument: I even have a (very few) recordings where pre-echo is noticeable but I think we're prone to over-estimating things like that versus universal support.

If the value of click-inducing wrongness wasn't such a well-understood force one would have to wonder how exactly all those writers were incentivized by owners of not-yet-free formats.

As it stands though, we just get to apply Occham: in online news media, A/Bing truth vs lies for clicks has a clear winner.

This is a symptom of PR copy paste article writing.

Exactly, they are all variations on Fraunhofer's termination announcement which of course recommends modern replacements instead of telling customers to use MP3 for free.

We are in a word where success as a write is churning out 2 fluff pieces of click bait rather than one research article.

The writers at all the AOL and Gawker franchises are the end result of this trend - a million monkeys sitting at typewriters.

Betcha it generated a lot of clicks.

tech "journalists" write about things they understand poorly. I'd love to have more insight into where the groupthink on this comes from though. maybe they're more familiar with covering other industries where an expired patent might mean that a pharma company would cease production of their brand name pill (and let generics sell it at lower margins instead)?

The did the classic lazy news thing of publishing a press release and with a new title. Nothing more than low value news filler.

Gizmodo has a history, in recent years, of this kind of ignorance. Something has changed in their hiring practices, it seems.

Not at all surreal when you keep in mind that the world turns on money. From that point of view MP3 is indeed dead.

Depends on whose money. For anyone who does not profit from those patents (>99.99% of the world) they'll save licence fees, which would mean more money for them.

I think his implication is that patentholders encourage the support/use of the format, abstract as it is (a concrete example being PC vendors obliged to sell windows machines only).

I see your point, but like the author I think mp3 has plenty of inertia even from this point in time. Even if I overestimate mp3 usage nowadays, those other journos are getting ahead of themselves.

> MP3 is really, really​ smart. Like seriously.

How so/any links or search terms for the aspects you're impressed by?

Was really blown away by [0], a description of the Fourier transform.

And also [1] is a good read.

[0] http://nautil.us/blog/the-math-trick-behind-mp3s-jpegs-and-h...

[1] http://ww.w.mpgedit.org/mpgedit/mp3.pdf

It's second nature to anyone with knowledge of signal processing. Sure Fourier transforms are very useful, but let's not be led to think that there was anything novel in the design of the MP3 format.

Let's also not forget that the orignal Frauhofer encoder was not exactly impressive in it's results. Anyone remember downloading 128k CBR files off napster, the warbled mess that was?

A lot of the advances came from the psycho-acoustic modeling and VBR algorithms from the LAME devs.

I remember having to download LAME binaries from South America, as recently as last month, and as long ago as 1999. And I also remember when it was just a patch to Fraunhofer's reference implementation, under a suspiciously shaky interpretation of the law.

I remember doing a comparison test way back in the late 90s between an officially licensed MP3 encoder and LAME on the same set of music WAV files, and was simply blown away when I could actually hear a difference in the results. Being the sort of anti-audiophile that I am, I then reduced the bitrate of the LAME encoding until I could just barely tell that it was too low, so that I could cram more music onto my hard drive, make playlists, and record them at 1x speed through the line-out port to an audio cassette tape, for my car stereo.

At the time, LAME made the smallest file sizes for the crap music quality that I tolerate. Nowadays, I encode at minimum of 96k/sec (sometimes better) just to annoy other people around me less, and to cut down on the suggestions that I get some real speakers for my phone. It ain't the speaker, folks; that's pretty much the best quality I can discern with my own ears.

When it comes to music, I think people pay far too much attention to audiophile opinions, and perhaps not enough to those who have physiological limitations on their enjoyment of music and/or recognition of speech-frequency sounds. But yeah, even I could tell when someone used a crappy encoder on their MP3 files. That's why I always ripped my own CDs, or specifically sought out lossless-encoded files, so that I could then use LAME to get an acceptable and consistent level of crappiness. Even now, I only download FLAC, and transcode my own MP3s for portable devices.

the real story is why several techblogs spun the expiration of a patent to be the death of the mp3 format. Clickbaity publishing out of control (has it ever been in control?).

Honestly I'm looking forward to Spotify running out of money and shutting down so this generation will wake up that they need to start duplicating and archiving their media before they lose it.

Have a strong feeling there is going to be a cultural black hole where large segments of music etc lost in the post-naptster/post-piratebay world because it only existed on the artists machine, Spotify's servers and YouTube's servers.

(I understand pirate bay is still kicking but its all certainly way more niche that it was 5-10 years ago)

I've gone the opposite route. After years of careful curating my music library, I started to feel chained to the past, listening to a handful of tracks over and over. Since music is connected to strong emotions, this would also bleed into other aspects of my life and cause me to be less forward-thinking.

I'm fully aware that Spotify and its competitors are deals with the Devil but if they allow me to feel less burdened, the price is worth it.

It's deeply hypocritical but I also secretly hope that at least a few people stick to archiving and curating just in case.

I trained a NI expert system on the kind of music I like, and turned it loose on the Internet, where it uses the Amazon wishlist API to make recommendations.

In other words, I had a kid, played my favorite music to the baby, and can now mooch off all the wonderful new CDs that show up in the house. All CDs get ripped and encoded as FLAC for the family media drive, and everyone transcodes their own lossy files for their own portable devices.

It's kind of an expensive solution, though, and occasionally fails to recommend music that I like.

Can you post code for this somewhere?

It has been around forever, and sufficiently detailed practical demonstrations are all over the network--often flagged with the acronym NSFW, which stands for "Natural Sapience Field Work".~

You left out the training system details for the neural network.

The NI is loaded into a biomechanical interface that provides sensory inputs, locomotive actuators, and environmental manipulators. Typically, only one researcher builds the entire device, and collaboration is not useful for that part of the project. (Unfortunately, as the network heavily exploits subtle implementation details in the mechanicals, it is very difficult to perform upgrades after the first stages of training are completed.)

From there, the researcher has to continually upload conceptual primitives through the sensory apparatus, and the NI prunes and rebalances its own neural network to establish basic foundation concepts such as object permanence, the acceleration of gravity, thermodynamics, ballistic path prediction, etc. Eventually, when the network is sufficiently trained, researchers may begin to input additional data through a natural language interface. Due to variations in the biomechanical devices, it is currently impossible to use standard bootstrap code to accelerate that process.

In order for the NI to be useful as a music recommendations engine, it is essential to expose it to music that you already like through its audio sensors, during the initial training phases. After approximately 8 years, the NI will begin to autonomously seek out music samples in the wild and recommend that you purchase copies of promising collections. The system is not perfect. It will occasionally issue recommendations for music that was already present in the training corpus, or for maliciously-formed music files designed to hack uninoculated NIs into recommending them. And it should be noted that these NIs have been known to abruptly diverge from preferences implied by the training corpus, producing wildly inaccurate recommendations thereafter.

It's probably just cheaper and more reliable to use an AI, but as long as this thing still works okay, I'm going to keep refueling it.

This is what happened to me as well.

I used to spend several hours per week on average looking for new music (new for me, might be old releases) and Spotify's AI basically does the job for me now. Well worth 10 bucks per month.

Also, the biggest factor in "sound quality" is the matching of the master and the device used to listen (most notably compression of the dynamic range, you want it strong on low-end speakers and/or noisy environments, and as dynamic as possible on high-end systems in good listening conditions). I'm very pleased with Spotify's masters, they're often equal or superior to the CD even, which is nice in an otherwise "loudness war" ("bricked") market (been that way sadly since the late 90's, which brings tears to many a sound engineer).

I can't speak for Apple Music (never tried it) but Google Play Music has absolutely terrible masters in many cases, way too compressed, sounding flat.

Note: I speak of dynamic range compression (audio technique), which has nothing to do with data compression (e.g. MP3). The actual end-user codec has actually little to nothing to do with it, it's all about mastering, i.e. targeted reproduction devices and listening conditions.

I embrace both.

I still manage an mp3 library and use a (modded) iPod because network access is unreliable for me sometimes. It's a curated library that I spend time adding to and culling from; it's its own little hobby.

I also really like streaming services. I like to be able to press a couple buttons and get a non-stop stream of music that I don't have to worry about, and can often lead to new artists, etc.

They each have their pros and cons; I see no reason to be dogmatic about either. The separation also helps with curation; I can keep social and casual music to the streaming stuff, and keep the library focused on 'dedicated' listening.

Thank you - my thoughts exactly. As an avid music collector and listener I understand that my interest in this issue probably isn't comparable to most other people, but I think there's no doubt that licensing deals will change/end and music starts to disappear from streaming services.

There are already and always have been large gaps in the streaming catalogs but those are mostly releases that have never been there in the first place. I guess it will be different when John Doe sees his Spotify library shrinking and realizes that he was just renting all this music.

Can I plug something to you guys?

I'm the same as you, I like collecting and listening to music, and I was tired at how fragile playlists are. If you move some files or rename them, they break.

So, I wrote a spec for a new format, which is resilient to pretty much all operations:


I'm currently writing a plugin for beets, but if anyone wants to help out with implementing an import/export plugin for a music player with library functionality, I would love that. I'm basically trying to scratch my own itch here, and hopefully one that many other people have.

You're right. Migrating playlist from one platform, or just a player, to another is huge problem. Keeping them in sync seems impossible.

A universal cross-platform playlist format seems like a much needed feature, but I don't se ehow this could work without significant changes in every supported player/software.

Here's a very real playlist issue I'm facing right now: I use Roon as my main audio player / music management solution. Roon manages my library which is located on a NAS. I have a few playlists created there, but no way to sync them to my iPhone unless I manage a separate playlist in iTunes which has access to the very same NAS library. It's not a deal breaker since I'm used to playlists being software-specific.

Unfortunately I don't see that changing anytime since everyone's answer to this problem seems to be using one single streaming service - a solution which, frankly, handles this particular problem very well, but doesn't work for me due to disadvantages of streaming discussed earlier in this thread.

The solution for that is what I do now: You keep one canonical copy of your playlists in UPL and use that to export to PLS or copy files to a directory, as needed. Then you import that playlist/files to your devices.

It would perhaps need an accompanying software service that could handle integration between music services you use. Almost like a music librarian.

"Have a strong feeling there is going to be a cultural black hole where large segments of music etc lost in the post-naptster/post-piratebay world because it only existed on the artists machine, Spotify's servers and YouTube's servers."

That has already happened several times. Anyone remember silent films? Most of the scores were never archived, so most of the original music that went to those films is lost to history.

That was also because a lot of those scores were actually performed live, but yes, it has happened already and it keeps happening every day - tons of books, movies and cds are never reprinted because they are copyright-encumbered but not popular enough, and eventually their master documents are misplaced or lost... if it can happen to Kubrick, Dr. Who and Joy Division, it can happen to anything and anyone.

I'm a happy Spotify user, but for discovery, not the actual playback.

Discover Weekly is highly helpful, and the Release Radar is basically functioning like my music RSS feed. I also use Spotify's save option as basically bookmarks on which albums to download.

Therefore, Spotify is still more than relevant to me, even though I'm keeping local copies of music. I can't really discover new stuff in my own library.

I'm a happy user of Spotify, but I have kept all of my MP3's and such locally. I am also well aware that Spotify may shut down some day, and I'll simply gather local copies of everything I like, during the shutdown period. It's not like they're going to shut down instantly from one day to the next, there'll be a reasonable warning period.

As for music disappearing, I have only ever seen that on my ISP's own streaming service, I've never actually seen it happen on Spotify.

Tracks get pulled from spotify regularly as deals expire and aren't renewed. My fiveour hour piano playlist is missing a few, as is my 400 song revising playlist from university (its down to 394) - what they are I don't know, I just know they are less.

It may not always work but there is an option in settings that shows these so-called "unavailable" tracks. Not helpful but it will at least bring closure.

I find that feature very helpful. If a track I like is no longer available and I enjoy it enough to want to download it, I can always go somewhere where it is available and buy it/download it. Much better than it just randomly disappearing and then having to remember what track could possibly be missing out of my 100+ song playlist.

Thanks for this! It is helpful as it allows me to make manual corrections

"there'll be a reasonable warning period" - wishful thinking, I'm afraid.

Do you really think they can just pull the plug instantly, on millions of paying subscribers? At the very least, there will be a 1-month warning period, to let paid subscriptions run out.

Besides that, I think it is in the best interest of the record labels to keep Spotify going. They're absolutely raking in the cash from streaming, and have just posted their largest industry revenue increase in 20 years. I think they would be willing to keep Spotify going on life support, as long as it makes them money.

> Do you really think they can just pull the plug instantly, on millions of paying subscribers?

Yes. Why not? There's no physical law forbidding it, and I'd wager their T&Cs contain a way to do so. They could even be magnanimous and refund the un-used portion of your last month, but you're still stuck without the music.

> I'll simply gather local copies of everything I like, during the shutdown period

Assuming their servers can handle millions of paying subscribes downloading all of their music all at once.

> Assuming their servers can handle millions of paying subscribes downloading all of their music all at once.

Not from Spotify, from other sources. A lot of stuff I listen to, I already have as CD rips. A lot of other stuff is available for sale on sites like Bandcamp.

We've seen enough examples, from the original dotcom bubble to present. "As of now, the service is no more, sorry."

Should they? No. Can they? Yes they can - why couldn't they?

Won't the local copies be DRM-protected in a way that makes them inaccessible when Spotify shuts down?

From non-Spotify sources, obviously.

A lot of the stuff can be bought on Bandcamp or through other sites.

I've long held that the great tragedy of the Information Age will be that nothing will be remembered.

Is it that nothing will be remembered, or that what will be remembered is a chaotic mess?

Or that everything will be remembered, but no one will remember what was important.

Even smaller bands still sell CD's or at least use Bandcamp.

I'm waiting for a mass movement of artists to offer direct sales at reasonable prices. I would love to own media files, but I currently don't even have the option of buying them in a sane (legal framework) and fair (to the artist) way at a reasonable price.

Until then I'm ok with streaming.

What about Bandcamp?

it's a good start but is still not artist direct, and is still not offering good prices

Vinyl FTW

Frankly, I'm not a fan.

I buy vinyl only as a last resort if it's a vinyl-only release or rare classic I'm interested in (yes, there's lots of stuff only available on vinyl - old and new). Then I just rip it and add the digital, lossless result to my collection.

The vinyl format itself is too inconvenient for my taste and the audio quality is too dependent on various factors like used equipment, condition of the media and quality of the pressing.

I like vinyl. But as a backup medium CDs are just as good if not more robust, I think. I have yet to experience a single CD failing. Even CD-Rs turned out to be pretty robust unless being handled with an utter lack of care.

>I have yet to experience a single CD failing

I have never seen a professionally pressed CD fail. I have seen lots of self-burned CDs fail after about three years. Of course there are archive-quality CD-Rs with projected lifespans of 100 or 300 years (if burned really slowly), but that's not what most people use for backups.

I have a pressed CD that failed. It is a regular audio CD from 1993 or 1994, and it developed tiny black flecks all over the aluminium foil. I guess it is not perfectly sealed. I think I still have it somewhere, even if I just bought a replacement for some € during a sale... so yeah, it can happen, after a small amount of years.

A small number of pressing plants produced bad discs during the late 80s and early 90s.


Main difference is not robustness. It's that CD are digital and can be copied perfectly.

CD-R, and especially CD-RW variants, are not robust, and not a good archival form. CDs themselves are pretty good.

Vinyl, with some care for environment, is really good for archival.

Worth knowing most new vinyl records come with a copy of the digital download too.

Amazon will also automatically add (most?) vinyl purchases to your Amazon Music Library, and allow download.


Hell no :)

The Amazon downloads seem to all be 44.1 khz mp3 (just downloaded an album of mine to check).

Wrong on the CD-Rs: I've seen them fail before, due simply to age. It happens with the cheap ones; the "archival quality" ones really do seem to be much better, but I have some ~15 year old cheapies that are dead. You can even see on the disc that the dye has changed colors and is uneven.

I've never seen a mass-produced CD (the aluminum kind) fail due to age rather than physical damage. However I have read about it happening, where the aluminum layer wasn't properly sealed in the polycarbonate and over time got corroded.

Vinyl is cute, and I love the huge cover art. But it's also fragile, takes up a ton of space and has compromised sound quality.

> AAC makes a lot of sense for low- and medium-quality applications where bandwidth is extremely limited or expensive, like phone calls and music-streaming services, or as sound for video, for which it’s the most widely supported format.

Nope; you can scratch "phone calls" from that list.

AAC (specifically, the AAC-LD variant) is not the best for low bitrate calls; you want a dedicated voice codec for that application.


AAC-LD is only geared toward voice in one parameter: frame size. It's basically just a "look, AAC can do this too if you want" feature.

Look at the remark there: "It can use a bit rate of 32 - 64kbit/s or higher". That's a whopping lot. 32 kbps is about the far-out upper bound on bit rate for using a voice codec. You can get very good call quality at half that.


Basically if you look at all the options for compressing speech in telephony, AAC doesn't make a whole lot of sense.

I thought Opus was technically fine for phone and able to compete on quality. IIRC it only suffers from lack of use. On paper it seems to be the universal codec for audio.

Opus is the default voice codec for Discord, and also available as a codec option for Teamspeak and Ventrilo servers.

Anecdotally, the sound quality is above and beyond what I get on my cell, without question.

Don't forget Mumble! Also Slack, apparently.

Its going to be hard to beat G.729's 6-8kbps bitrate. I can barely hear the difference between G.729 and G.711 on our voip system.

Interestingly the G.729 patent pool also expired this year:


> since Licensors have agreed to license the unexpired Licensed Copyrights and Licensed Patents of the G.729 Consortium Patent License Agreement under the existing terms on a royalty-free basis starting January 1, 2017.

Am I reading that right? The consortium has graciously begun issuing royalty-free licenses now that the patent has expired?

How nice of them.

AAC-LD and AAC-ELD get plenty of use as the preferred codec for Facetime on Apple products (last I checked).

MP3 isn't dead, just journalism

The facts are true, the news is fake.

This is the truest statement in the whole thread.

Yep. To take "MP3 encoding is now free for anyone to use!" and turn it into "MP3 is dead!" is one of the worst cases of bad reporting this year.

MP3 is now as dead as the GIF. Why, I haven't seen a GIF for at least... 15 seconds.

At any rate, (pun intended), I'm perfectly happy with my vorbis files and not in any hurry to convert to MP3. But I'm glad I don't have to worry about the silly patent any longer.

It bugs me that "reputable" news outlets like Fortune Magazine were running stories which essentially amounted to a Fraunhofer press release with a couple of paragraphs tacked onto it.

Clearly there's absolutely no effort involved in being a journalist anymore.

> It bugs me that "reputable" news outlets like Fortune Magazine were running stories which essentially amounted to a Fraunhofer

You skipped a chapter on the history of the web. Formerly "reputable" outlets have transitioned to rubberstamping their brand on any blogger they think can help them preserve their business model, so while they might admittedly still be frontrunners in name recognition and still have credibility in narrower contexts, they have all but morphed into full-time clickbait farms.

One has to wonder if its actually just an AI doing web scraps looking for 'news' and posting a summary or just direct excerpts from the source. Then changing/adding a few things it knows are clickbait/headliners.

If this isn't being done, it seems like something that could potentially be achieved with current tech, if not now, then in the near future.

Journalism is dead.

Full length stories and analysis are still very much in demand, but the actual news communication aspect of the profession is indeed doomed.

I wrote a similar article in Arabic.

yes, because MP3 is now similar to public domain it does not mean it's dead. only the business of patent trolls behind it is dead.


There are many examples of patent trolls but this isn't one. MP3 was new, innovative, and the result of long effort. The patent fees directly funded new research of even better technology. That's the patent system working as designed.

I just read your blog post through Google Translate. You seem to claim that they patented the use of Fourier transform for storage of music. This is not what happened at all. The patent was granted for specifying which part of the Fourier transform can safely be discarded for storage (and reconstructed while playback) without harming the sound you hear too much. That's not something they just looked up in a 200-year-old book but took more than a decade of complex research. The result of this research is their actual product and it's a highly valuable one.

I'm a person who oppose all kind of software patents. any use of software patents is a patent troll (if you have patent, you should have a clear patent policy that it's for defensive purpose).

According to the case of Bilski, patents should involve a machine, otherwise it's invalid. But trolls don't care of their patents are invalid, because of how patent cases are handled in courts


Your political stance doesn't change that what you wrote is simply wrong. They did not get a patent for the Fourier transform but for something completely new which no one thought of before.

Fraunhofer would not have developed audio compression if they couldn't earn money with it (they are a non-profit by the way and the money earned funds new research). That's their business model. They have novel ideas, do hard work to make them possible, and sell that to companies. Are you suggesting people shouldn't get paid to do their work?

Algorithms are the machines of the 21st century.

The court rule in Bilski case, does not agree with you. You can't patent math. It's not my political stance, it's the law. You are not supposed to patent formulas.

The Fraunhofer Society is not and never was a "patent troll". They do actual research and one of their income sources is licensing the resulting patents.

Fraunhofer put the source online without a license at the ISO site, let the community convince itself that MP3 was a standard and therefore unpatented, let the community write all the encoders and decoders and surrounding tools for years, then turned around and asked everyone who made their file format popular for 10K USD.

Troll is a gentle word, the author of ffmpeg describe patents of software as "gangsters asking for protection money". All software patents are bad see my reply above


It's not really a business of patent trolls.

It's the unfortunate effect that the German research organisation that developed it only finances all their research with patents.

Related: I haven't heard whether popular free software such as Audacity, Linux distros, etc. will begin including LAME binaries by default as a result of the patent expiry. Anyone know if such plans exist?

well, for one example, have a look at http://www.slackware.com/changelog/current.php?cpu=x86_64 - specifically

    Sat May 6 23:12:02 UTC 2017
    a/glibc-solibs-2.25-x86_64-2.txz: Rebuilt.
    ap/cdrdao-1.2.3-x86_64-3.txz: Rebuilt.
           Recompiled to support libmp3lame.
    ap/sox-14.4.2-x86_64-4.txz: Rebuilt.
           Recompiled to support libmp3lame.
    d/flex-2.6.4-x86_64-1.txz: Upgraded.
    kde/k3b-2.0.3-x86_64-3.txz: Rebuilt.
           Patched to build with ffmpeg3 and gcc7.
    l/ffmpeg-3.3-x86_64-1.txz: Upgraded.
           Recompiled to support libmp3lame.
    l/glibc-2.25-x86_64-2.txz: Rebuilt.
           Reverted a patch that causes IFUNC errors to be emitted.
    l/glibc-i18n-2.25-x86_64-2.txz: Rebuilt.
    l/glibc-profile-2.25-x86_64-2.txz: Rebuilt.
    l/gst-plugins-base-1.12.0-x86_64-1.txz: Upgraded.
    l/gst-plugins-good-1.12.0-x86_64-1.txz: Upgraded.
    l/gst-plugins-libav-1.12.0-x86_64-1.txz: Upgraded.
    l/gstreamer-1.12.0-x86_64-1.txz: Upgraded.
    l/lame-3.99.5-x86_64-1.txz: Added.
    xap/MPlayer-1.3_20170208-x86_64-4.txz: Rebuilt.
           Recompiled to support libmp3lame.
    xap/audacious-plugins-3.8.2-x86_64-3.txz: Rebuilt.
           Recompiled to support libmp3lame.

I recently installed Audacity on a new machine and was pleasantly surprised to not have to poke around and drop the LAME libs manually for it to work with MP3.

Linux distro's are starting doing it right now.

Not sure about Audacity, but I guess that will be the same.

A big argument for using lossless instead of lossy are Bluetooth headphones. When you're listening through them, you're essentially re-encoding one lossy format into another lossy format, which degrades the quality much more than each of the formats does by itself.

I simply refuse to use Bluetooth headphones. Problem solved.

It's not dead, although as storage prices continue to decline, one wonders why you would still compress audio you care about with a lossy codec.

The file size increase will continue to be an issue for a long time. As hard drive space started becoming cheap enough for PCs, people started wanting their songs on phones and MP3 players, where the storage was still expensive. As phones are now starting to get cheap enough internal storage for the flac file size not to matter, people are starting to stream music, where the file size is again important, as people are streaming music through cellular, where the price per gigabyte is still high.

Maybe, once most people, even in the third world, have unlimited data plans, music file sizes will stop mattering. Or maybe we will have moved on to something else entirely where big files is still an issue. Maybe people will want to stream music from earth to space stations or lunar bases, I don't know.

I think storage on phones is a huge factor right now. A lot of the world is probably using phones with 8gb and 16gb of storage. That probably becomes ~1gb and ~9gb of free space after the OS. I'm not sure how much those countries influence the media format but even in the US a lot of people I know still have 16gb storage on their smartphones.

"Storage prices" are abstract; my phone's non-extendable 32 GB of storage are real. Compressing my library from iTunes AAC to lower-bitrate Opus made it more than two times smaller — down to 5 GB from 11.

I recall an offhand comment from a source I can't remember¹ that some people when presented with mp3s encoded at various bitrates and flac encodings of the same song preferred low-bitrate mp3 above flac above high-bitrate mp3. The person theorised that it was because the low-bitrate encoding had more noise, which gave the sound additional texture. I draw an analogy with dithered palette images which can look different, but better, than full-colour images.

¹ Yes, this is the ultimate in unreliable anecdote, sorry.

I wouldn't be surprised if people prefer the low bitrate MP3 songs because that's how they've always heard it. Most people have never heard "good" audio.

Your analogy is quite apt.

Because you literally can't tell the difference.

It's not about audio quality alone, it's about collecting a flawless source.

If you want or need your library in another format at some point you really should not transcode from a lossy source. You may say that won't ever happen, but storage is cheap it's very short sighted to risk being locked into one format. It's not something you could easily fix later.

I see it as one reason why lossy won't go away: reencoding losses are the tape-to-tape copy of the digital age, the very big M, very small V MVP of DRM: if you can't keep consumers from copying, the least you can do is keeping the master copy to yourself.

> ...the least you can do is keeping the master copy to yourself.

Except that lossless audio is far from hidden and widely available as soon as you look past the iTunes store. I buy my music in a lossless digital format exclusively and frequently - from multiple stores, without any DRM and with no problems at all finding it. If the music industry tried to lock the lossless source away, I'd say it does a very bad job.

>- from multiple stores

Could you give an example of some?

- Bandcamp

- Boomkat

- Qobuz

- Bleep

- Artist and label stores


There are others like HDTracks, but they always seem to be selling snake oil to audiophiles. Just my impression. I don't see any point in paying extra for more than CD-quality. I draw the line at 16 Bit / 44.1 KHz.

HDTracks is an example site that often has lossless for sale.

Most importantly, audio CDs still use uncompressed WAV. Buying and ripping the audio CD for a popular album is a pretty surefire method for getting a lossless copy, even if it's not always the easiest.

It might be short-sighted if the format doesn't have excellent open source decoders. MP3 won't die ever, because of LAME.

But you can covert to whatever lossless format you need without worry. Lossless to lossless conversion is possible an infinite number of times (theoretically), vs lossy to lossy you get maybe a couple for free.

As long as you keep the MP3s around, converting to any format for the 100th time is no lossier than the first time. What you need to avoid are chains of lossy formats.

Once upon a time I converted 320kbps MP3s of Dark Side of the Moon to 320kbps OGGs. It ruined a certain cymbal sound.

Never convert between lossy formats, not even once.

It likely won't, that's true. My point remains though: With storage that cheap and lossless audio sources widely available, why would you not use the lossless option?

Why not both?

MP3 is more widely supported than any lossless audio codec. Lossless compression enables the collector to keep an original.

Not everyone is a collector though - for them, MP3 is probably fine, and if they need a different format later they'll probably just stream, re-encode, or re-purchase, in order of likelihood.

Maybe LAME can be renamed to LIME now?

>It's not about audio quality alone, it's about collecting a flawless source.

Most people don't care about that

Everything's going to have a format, including your "flawless source". ...and you're not locked into anything with MP3.

You are unless you're fine with decreasing the quality of your files. While you can convert lossless to lossless as often as you want without any degradation, each time you convert a lossy source to another lossy format it'll decrease audio quality, similar to repeated JPEG conversion.

So long as it's high bitrate (like 320k), then most likely.

If I play only 128k mp3s on a modern hi-fi system, I can tell the difference. I prefer FLAC myself. It's a little more space, but at least I know I have a real archive of a CD.

I have rips of some good bands bands that never made it past the bar scene and who abandoned their MySpace pages long ago, so it's nice to know I have a preserved copy I can share with other audio-nerds.

I rip all my files to flac, mostly because I don't know which lossy formats will be popular in 4 years.

Anyway, anecdote time:

I was actually part of a double blind test.

On a really really good setup with very little background noise I can tell that there is a difference at 320kbit (encoded using LAME) with about 65% reliablity for short pieces of music, which was the second best in the group.

I could not, however, say which sample was better.

I dare you do blind tests! :) I also think the fact that the hi-fi system is modern has little to do with it.

You really don't need blind testing to tell that 128kbps MP3s are awful. Certain common instruments will produce blatant artifacting all over the place.

128 is just barely good enough to be tolerable for a lot of people, but nobody would claim it's transparent. AAC does a lot better at that bitrate.

128 is just barely good enough to be tolerable for a lot of people

As someone who grew up loving music while listing to Nth generation tape copies on a really shitty walkman I just don't understand this line of reasoning.

Same here. Also, I now have tinnitus and partial hearing loss, so the audio snobbery makes me incredibly envious of those people that could always hear so much better than I can, such that they just have to complain about hearing the little extra things (or notice the missing bits) that annoy them.

It's great that audiophiles can hear so well, but when they start talking about tubes and premium cables and vinyl records and gold-plated contacts, I just want to give them an otitis media infection that cuts them down to my level. You just don't know what you've got 'til it's gone. I know that's wrong to wish such ill upon others, but it's so much harder to suppress my envy when I'm being inundated by what appears from my perspective to be superstitious nonsense.

As someone who can still somehow enjoy music when it's encoded as a 64k/s MP3 file, I am baffled as to how someone can say the very same music can be bad at 128k/s. I'm just glad that I can still hear it at all!

Depends. If the system more modern than actually hi-fi then chances are high that there is so much hidden DSP signal mangling that all the occlusions and illusions so carefully balanced by the encoder don't work half as good as expected.

Absolutely this. mp3s wreak havoc with Pro Logic II decoders (stereo->surround processors), regardless of the bitrate, by stripping phase information which the DSP uses for sound steering. Also, the extra speakers allow you to spatially isolate sounds that would normally be masked - this can actually be a richer listening experience, but with lossy encoding it's a mess.

More generally, this applies to any scenario where the sound is distorted or altered by something. A crappy speaker might have a perverse frequency response. A room reflection might manifest as a comb filter. Both will destroy the careful masking assumptions of any psychoacoustic model.

Lossy compression should be avoided as an archival format whenever possible, even if you think you know better, because you never know when you might need the information you thoughtlessly threw away.

MP3 isn't transparent. The pre-echo artifact is inherent and unavoidable no matter how high a bitrate you use.

And even if it were, that doesn't make it worthwhile. Gzip compresses plain text by a factor of 4 or 8 and is lossless in the absolute sense, but most of us don't keep our text files in .gz form.

Transparency is relative to the user. In the 192/224kbps range (with a good encoder), only a very small part of the population is able to detect artifacts in a blind test.

Text files are not a good analogy, as they have a different workflow compared to musical files.

There are filesystems that transparently compress large text files. They're not in common use because CPU/disk IO is more expensive than disk space, and filesystem compression has fairly low yield on a standard desktop.

Compression also means less disk I/O, not just space.

I use ZFS and NTFS compression on all the things. For example, ~/.android on my desktop: "Size 1.49 GB, Size on disk 293 MB". Not everything compresses as nicely but in general you can save a lot of space on a desktop.

How do you check that for a single directory? I only know how to check for a filesystem or pool.

Yeah I don't know how to check that on ZFS. That result was from NTFS on Windows, I thought the phrase "Size on disk" was enough for everyone to tell it's Windows :D

It's gaining popularity in modern file systems as CPUs paired with fast compression algorithms like lz4 or lzo are generally fast enough nowadays to easily outperform most common storage devices. Saving space also means you need to read or write less form/to the often "slow" medium which makes things faster.

Most of us don't have several hundred Gigabytes of textfiles though.

Most ;)

Well, not the first listen. But after MP3s aren't supported anymore and you have to convert it to another lossy format you might start hearing the difference. The point of lossless formats are that you actually own an original. That you have a copy you should be able to convert to anything else no matter what the market trends.

> But after MP3s aren't supported anymore

And how many decades from now do you expect that to be the situation?

You know what happened last time someone said that.

No. Please enlighten me :)

Y2k. I get what you're saying, MP3 is going to be around for a long time, but I really don't this whole craze of leaving future-proofing for the future is a good idea.

Modern Linux machines still ship with FS drivers for Amiga filesystems, and have software to play MOD-files from the early 90s, despite nobody creating new music in that sequencer-format any more.

I think MP3 and us are going to be fine for quite a while. This is after, all not a bug that needs to be fixed.

Basically I don't think about this is a "problem" and this I don't agree about the "pushing it ahead of us" metaphor.

I'm not sure that I care all that much when my original lossless source is some poor quality master tapes or vinyl that somebody has digitized from the 60s or 70s.

I can tell the difference. And if used proper equipment you would too.

Not everyone is an audiophile.

If I want to store some really high quality audio, I'd use something like FLAC. But for general listening? Nah, MP3-320 is fine.

V0 has replaced 320 for a while now

I can't fit a fraction of my collection on my mobile devices in a compressed form. Or my laptop. Moving to Flac would fill up almost my entire MacBook SSD.

I do because music collections still take large amounts of space in lossless formats.

From another point of view, why would you care about compatibility across device if you are having control on what you are listening on.

MP3 has been "dead" to me ever since the non patent encumbered formats started becoming popular. Mind you, I've been using Linux as my primary desktop OS since the mid 90s... So I have a vested interest in open formats.

These days, all my music is in a lossless format anyway. Especially now that my phone has enough storage space for it.

Anyhow. I'd say that if anything... The patents expiring way the heck sooner would have been healthy for the format.

I think the contrary, I can see MP3 getting even more popular now that it has no patents.

Not sure how the author comes to the conclusion that opus isn't widely supported.

Opus is supported in all major webbrowsers and natively in modern Android systems. I'd call that widely supported.

> Opus is supported in all major webbrowsers and natively in modern Android systems. I'd call that widely supported.


Not supported in Safari. I guess the reason is that iOS devices don't have hardware encoder for it and Apple doesn't want to compromise battery life (maybe marginal issue with audio, but big issue with webm).

> I guess the reason is that iOS devices don't have hardware encoder for it and Apple doesn't want to compromise battery life (maybe marginal issue with audio, but big issue with webm).

I rather think that Apple doesn't like the idea of royalty-free codecs which would make it easier for free operation systems to support multimedia on the web. AAC helps to keep you locked into macOS or Windows, because it works out-of-the-box there (in contrast to Linux distributions like Fedora).

On the other hand, the only audio format I'm aware of Apple inventing in the last 20 years is royalty-free and available under the Apache license.


It wasn't royalty-free from the beginning, in contrast to FLAC. Guess which browsers don't support FLAC: http://caniuse.com/#search=flac

Opus is not supported in IE11 and Safari (both iOS and Mac). If you take the 'global' statistics from caniuse.com that's about 37% of all people.

I wouldn't call a technology that's only available to 63% of your users 'widely supported'.

I guess it's relative. Compared to MP3, a lot of stuff could be considered "not widely supported".

AAC was introduced in 1997, does any one know when will its patents expire as well?

And we haven't had any improvement in Audio compression since then. MP3 - AAC and That is it. All the others are at best AAC similar quality / bitrate.

At 256Kbps, the majority couldn't hear a difference between MP3 and AAC. At 128Kbps it is only slightly better.

We dont have anything like HEVC which is an order of magnitude better then, say MPEG-2 at low bitrate.

Disagree with the author on the technical merits that AAC only sounds marginally better than MP3 at 128kbps and higher. For certain audio shapes, MP3 is a rather bad™ compression (as is, to some extent JPEG for some image shapes), whereas AAC produces much better results due to different compression mechanics. Much like MPEG4 ASP and MPEG4 AVC and HEVC produce different compression artifacts, with ASP having much worse artifacts.

> For certain audio shapes, MP3 is a rather bad™ compression (as is, to some extent JPEG for some image shapes), whereas AAC produces much better results due to different compression mechanics.

Do you base this assertion on blind testing? I'd really like to know how much of the people that says "MP3 sounds X/Y/Z" actually did one (actually, I'd like all of them to actually do one), because in the blind tests I've participated/seen participating, with a modern encoder and mid/high bitrate (in the average range of 192/224 kbits) users were systematically not able to hear any difference.

Of course I found some exceptions; for example, a friend of mine had good hearing on high frequencies, therefore, he could immediately spot 128 kbps CBR mp3s which have a lowpass threshold at ~16 khz.

I think I fall within that latter category. I have done tests myself, in an environment I feel relaxed in and comfortable, and have been able to hear the difference. I also hear the difference going a step above, with lossless.

You know what needs to be resurrected? MPEG1 Video Codec. It scales to 4k x 4k and very soon to be patent free.

There is a bunch of early-internet tech that will soon get out of patent locks, it could lead to a new Golden Age.

Well, it's not not patent incumbered, but there is no need to use it for anything besides playback of existing legacy media that doesn't have a lossless original. Otherwise just use Opus for playback purposes.

As a former cakewalk software user it used to piss me off royally that I had to pay for the MP3 license to unlock the feature. It was a trivial amount (19bucks?) and I understand why they thought it was a good idea(their way of sticking it to Fraunhofer, and a ridiculous fee) but it ended up feeling like the customer that just spent 500 bucks on software was the one getting screwed especially considering how much cross licensing cakewalk did/does.

Tangentially related: Being out of touch with the podcast scene for quite a while I was interested to follow the 'Overcast' link. Unfortunately that turned out to be for platforms I don't own.

If there are some passionate podcast listeners here,

1) do you have recommendations for an alternative to Overcast (ideally cross-platform, Android/Linux required, Windows desirable)?

2) can you recommend HN related (overlapping with the content here) podcasts?

I'm looking at the Tunequest list at http://www.tunequest.org/a-big-list-of-mp3-patents/20070226/, and it looks to me like patents #5,924,060 and #5,703,999 (the last two listed) shouldn't expire for a few months yet. Is that list mistaken?

AAC was released in 1997, why they start to advocate AAC after MP3's patent is expired?

I thought they state "MP3 is dead" just because its patent is expired, while AAC isn't:


Long live the MIDI!

TIL MP3 is only one year younger than JPEG, or considering i was using JPEG daily years before MP3 took off, I remember Winamp version 1

Finally, LAME can become LIME.

Can the Opus codec be emulated with asmjs in Safari or iOS?

You may be interested in the following project from the Wikimedia Foundation:


Frauhofer announcement did not mention expiry of patents.

So many people here do not want to face the reality that MP3 is dead. I know, my dad who worked in the MP3 factory for 20 years was laid off last week just after the last MP3 came off the assembly line. I suspect the more entrepreneurial amongst you here may want to start scouring peoples Recycling Bins for old MP3s that you can break down into replacement parts to sell to collectors.

Netcraft confirms it, MP3 is dead.

By mp3, do you mean .mp3, the audio file format, or an mp3 player ? Because the way I understand it, the article talks about the file format, not the hardware.


I think it's a joke

It's odd how this has hit the frontpage, but hasn't had a single comment on it yet.

Now it has two.

I upvoted it because of the annoying articles and comments that came up regarding this. Going from the "I can hear the difference" to "It's dead now".

This article points the facts out clearly and I don't see anything to discuss about there.

MP3 isn't dead :)

It's just your typical blog; not saying anything very new or interesting.

And yet it's on the frontpage. Like I said: odd.

> AAC and other newer audio codecs can produce better quality than MP3, but the difference is only significant at low bitrates. At about 128 kbps or greater, the differences between MP3 and other codecs are very unlikely to be noticed, so it isn’t meaningfully better for personal music collections.

This is patently false unless you listen on bad audio gear (bad DAC, bad cans).

I don't even have serious equipment (MBP+iPhone so guess about the DAC, as for headphones: AKG k514 mkII, Sennheiser Momentum (over ear), SoundMagic E10 (intra), Marshall Monitor (not mine), BeyerDynamic DT 770 250Ω (not mine)) and the difference is downright obvious going from 128k to 256k almost whatever the place and definitely noticeable on select features from 256k to 320k.

Some corner-case audio patterns are just known to make MP3 encoders choke whatever the bitrate (a higher bitrate merely mitigates the problem for MP3), while they pass with flying colours on AAC. Sadly those corner cases are not just theoretical and have a real effect for some songs. At 320k though it won't make the song any less enjoyable (which is another metric entirely) but it can definitely be noticed (which is what the article claims).

> This is patently false unless you listen on bad audio gear (bad DAC, bad cans).

Double blind tests suggest this is true for somewhere between 128 kbit and 256 kbit (256 kbit is CD quality, while 192 kbit is near CD quality, and the difference between 128 kbit and 192 kbit is miles compared to 192 kbit and 256 kbit; hence I suggest 192+ kbit. However, its 2017 and 320 kbit or lossless isn't a huge issue anymore; its even OK over 4G). Also, my quote is from 2000; ever since we swapped to LAME and ABR > VBR > CBR; ABR isn't even a static (constant) bitrate.

Quoting r3mix.net from 2000:


128 kbit/s is not cd quality

256 kbit/s is cd quality (x) (in case of Lame or some Fraunhofer, not Xing)

In february 2000 c't magazin organised a blind listening test. 300 Audiophiles were involved, finalists tested 17 1-min clips from different artists (classic and pop):

original CD recording

128 Kbit/s Joint Stereo [MusicMatch (FhG) v4.4] encoded PC decoded Mac

256 Kbit/s Joint Stereo [MusicMatch (FhG) v4.4] encoded PC decoded Mac

all on cdrs and played in a Recording Studio on:

B&W Nautilus 803, Marantz CD14 with amp PM14 (Straightwire Pro cabling and extra's) [DM30000- so bit more than $15000]

Sennheiser Orpheus Electrostatic Reference-headphones with tweaked accompanying amp (digital and analog out) [>$10000]


90% of the 128 Kbit material was picked out

MP3@256 was rated to have the same music quality as cd!

If you find MP3@256 to be of inferior quality compared to the original cd, you're very likely to be doing something wrong with the test (correct decoder, no objective double blind testing, DSP filters distorting the process, ...) Maybe this is something for you. You can always read the article in the german c't 6/2000 on p92.

The treshold of mp3 transparency lies somewhere between 128kbit/s and 256kbit/s, depending on the kind of music and your hearing and equipment.

English translation of the c't article: https://web.archive.org/web/20140217151727/http://www.geocit...

"If you find MP3@256 to be of inferior quality compared to the original cd, you're very likely to be doing something wrong with the test"

Try running that 256kbit MP3 through a pitch shifter. The quality loss becomes very apparent very quickly.

Listening through a pitch shifter is wrong, unless you are evaluating the codec for the purpose of listening through a pitch shifter. That's the thing - in normal use, high bitrate MP3 still sounds "the same enough" a lot of the time.

"Try playing a random binary blob as a video file, it might play, it might not"

Mp3 is a clever compression algorithm made for consumption. It isn't intended to be pitch shifted...

FLAC claims to be lossless/archival yet the same artifacting appears when pitch-shifted.

The author claimed that codec choice matters little >128k, not that 128k is of equal SQ to 256k.

BTW, if you can easily discern 256k from 320k, I'd venture to guess that you are a 1% "golden ear" listener. OTOH, even my 60yo ears can tell 128k from 256k.

I understood that but didn't made it point clear: my point is that it matters in a bit more nuanced way. I'll use [0] as some sort of visual support to help transcribe how I perceive it:

- 128k MP3 is like watching a Q=10 JPEG (artefacts and blockiness all over the place) while 128k AAC is much more like Q=50 (artefacting is much much more local, like the little grass patch on the left side of the stone)

- 256k MP3 is generally great but still has artefacts happen occasionally in some local, specific areas, similar to Q=50 (but not in the same way as 128k AAC), and definitely not on all songs, while 256k AAC is very close to perfect in any situation.

- 320k MP3 is very close to perfect save from very specific (and rare!) pathological corner cases that throws encoders off (like you have to have some specific bass line and some footwork on the bass drum and some very specific cymbal then the cymbal gets subtly crashy and noisy while the bass work gets subtly glitchy and muddled), which doesn't happen with AAC. Yes I've double blind-tested this with a friend who did not believe me, and used multiple MP3 encoders for that.

So to sum things up:

          MP3        AAC
    128k  annoying   enjoyable
    256k  excellent  awesome
    320k  superb
Those being ordered this way: annoying < enjoyable < excellent < superb < awesome

It's not much that I have some "golden ear" that I've been trained into looking for some patterns when I was studying DSP and now that I know what to look for they just annoyingly pop out. It's a bit like watching logs fly by just fast enough that you can't read them yet spot some oddity and just go "wait, there was something right there", and should you scroll back, sure enough, here's the deal standing out like a sore thumb.

Taking a bet that audio equipment gets better and cheaper with time I'd rather have a reasonable quality as a source, especially in light to the storage space and general bandwidth we have at our disposal. Then again in an increasingly streaming world we don't get much in the way of a choice whether it's about formats themselves or encoders.

[0]: https://en.wikipedia.org/wiki/JPEG#Sample_photographs

In a 2014 listening test, 136kbps MP3 was generally rated as somewhere between "Perceptible, but not annoying" and "Imperceptible": http://listening-test.coresv.net/results.htm

It really does depend on the music you're listening to IME. Some stuff sounds ok on 128k, but a lot of the electronic music I listen to is horrendous on 128k. Which is a real shame as a lot of sets I have I can only find in 128k mp3.

Spot on. This is something that is rarely mentioned, but having an extremely varied taste in music - I've noticed this exact thing.


I have never seen anyone pass that bullshit-detecting test. 128kbps encoded with lame is barely detectable, and anything higher just isn't.

This test is relative to the sound system and listening environment in use at the time of testing. It's not useful as support for absolute statements like "128kbps encoded with lame is barely detectable, and anything higher just isn't."

I'm not a snake oil audiophile, but it's obvious that conducting the test on a train with crappy $1 earbuds and an old Realtek DAC will yield results that are considerably different from e.g. http://www.jeremykipnis.com/Jeremy_Kipnis_-_Home_Page.html

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact