1. This is an 2008 article. Per Guidelines you should put years in the Title.
2. MP3 has improved a lot over its lifetime. LAME was already used for default by year 2000. When people say MP3 was good enough, they refer to MP3 encoded with LAME. ( Rant: When we people learn the codec, encoder and the encoded results are different things? 2023 and I see this mistakes everywhere still )
3. Even iTunes AAC has seen lots improvement since 2008. Especially in the 256Kbps+ Range.
4. And when AAC is mentioned. That is AAC-LC ( Or AAC Main Profile which isn't all that different ). AAC-LC ( Low Complexity ) has been declared as Patent free by RedHat. There is no reason to use MP3 today.
5. The definition of "CD-quality" alike went from MP3 128Kbps to now AAC 256Kbps. And arguably that is true for consumer market. Even Hydrogen audio has repeated these test multiple times.
6. I still prefer the codec MPC, Musepack (https://www.musepack.net). Sorry I just had to write it out. Sadly it never gained any traction.
7. If we have to be picky about frequency range, may be CD itself isn't good enough and we could use SACD?
8. Lossless is making a come back. Storage and Bandwidth cost continues to fall. ( Arguably not true for NAND, but let's ignore that part for now )
9. It is ironic when Lossless could gain and be used mainstream, Wireless earphones are replacing traditional earphones. Meaning your music will be re-encoded before it is sent to your earphone. And No. Most Android or iPhone dont have AAC pass through. i.e Your AAC encoded files will still be re-encoded before sending it your bluetooth earphone.
Why is the standard considered to be CD quality? In that way, the article shows its age. Today you wouldn't be talking about 44.1kHz 16 bit, it would be all about 24 bit 192kHz. If you're looking at spectrum plots, CD is very much on the low quality side of the spectrum of what's possible. Maybe we should be considering megahertz sampling rates and 32 bits, surely we have enough bandwidth.
Why not then? Because there is a ton of science and empirical evidence that humans cannot hear the difference[1]. Good engineering is about meeting the requirements with minimal cost. If the requirement is that it sounds good to humans, and the cost is number of bits to encode (and thus store and transmit) the signal, then modern codecs like Opus are clearly superior to uncompressed and losslessly compressed signals, much less higher sampling rates.
If your goal is something other than good engineering, for example the aesthetic satisfaction that the bits are the same as what the mastering engineer put on the CD, or for some reason caring how clean spectrum plots of artificial signals look, then the arguments may have some merit. But let's be clear on the goals.
The real issue nowadays is that you shouldn't record or master at 44.1kHz (or multiples) and go for 48kHz instead.
But not because it sounds better. Simply because 48kHz is what your computer and phone natively clocks its audio codecs at. It's done because that's generally an integer fraction, but the why doesn't matter as much as the fact itself; PC "HD Audio" and phone codecs are 48kHz.
Yes you can resample, and yes you can resample without it being audibly noticeable. But it's an extra step where you're at the mercy of whoever implements it to do it right. Doing it wrong may also include noticeable delay, breaking e.g. A-V / lipsync.
Can you build HiFi systems that support 44.1kHz and maybe dynamically switch their clock source as needed? Sure. But what's the easiest way to build a HiFi system these days? You just stick an off-the-shelf embedded device in it, which likely uses standard PC/phone tech…
So just ship 48kHz.
(Similar argument for video recording in 50Hz countries btw - unless you are recording for TV/broadcast, you should always shoot at 60 / 59.94fps. Because that's what PC and phone screens run at…)
everyone should really watch https://xiph.org/video/vid1.shtml and https://xiph.org/video/vid2.shtml
Utterly useless. No listening tests except for a custom made sound file which is designed as an artificial worst case for the codecs. You might as well benchmark a text codec on /dev/random...
> We recommend that, for serious listening, our readers use uncompressed audio file formats, such as WAV or AIF—or, if file size is an issue because of limited hard-drive space, use a lossless format such as FLAC or ALC.
I recommend that, for serious listening (for some weird definition of "serious"), go to a music concert. PCM is also a lossy compression due to the quantization step, albeit its effect is much less pronounced for so many reasons that no one even thinks it as a "compression" method. If you can tolerate PCM, you should be also able to accept some good enough lossy codecs---I don't know if that includes MP3 or AAC or Vorbis or Opus or whatever, though.
Let's assume there are people who are able to hear a difference. Why does it matter to a majority of people and the way they consume music? Maybe I'm from a spoiled generation, growing up listening to FM radio and tapes, even copying from tape to tape.
A lot of rock music lives from the imperfection of audio equipment, people spend a considerable amount of time replicating the behavior of vacuum tubes. Even techo producers like Robert Babicz record to analogue tape machine to enhance the final result.
Their recommendation doesn't make any sense. They first explain that lossless compression reproduces exactly the same data as when uncompressed:
"Lossless compression is benign in its effect on the music. It is akin to LHA or WinZip computer data crunchers in packing the data more efficiently on the disk, but the data you read out are the same as went in."
...but then recommend uncompressed over lossless compression for "serious listening":
"We recommend that, for serious listening, our readers use uncompressed audio file formats, such as WAV or AIF—or, if file size is an issue because of limited hard-drive space, use a lossless format such as FLAC or ALC."
OP forgot to put (2008) at the end of this.
Today's standard isn't "CD Quality" anymore. There is literally no audible difference between MP3 320kbps, which covers the complete range of human hearing up to 22kHz and FLAC which covers all the way to 192kHz, which is lossless. At this point digital audio has surpassed what the human ear is capable of hearing, and any advancements to this is superfluous as far as music is concerned.
The only advantage to raw or lossless formats for music is archiving, as FLAC can be converted into other formats without incurring additional quality loss. For listening, it is now more important to have good equipment rather than a lossless format, and for streaming it is generally preferable to keep bandwidth requirements down.
The only reason I can imagine to continue expanding the capabilities of lossless audio is for scientific purposes and machine learning where the limits of human sensory perception isn't a limiting factor.
With a headline like that it felt like 2002 again.
The article's byline has 2008.
A 2023 update could be interesting comparing the streaming providers' choices, and persistence of choices, now that monthly subscriptions, rather than actually owning anything, are so dominant.
What's really infuriating to me is that, at one point, Apple took a reasonable stand with their music streaming and said "256kbps AAC is CD quality" (it is). And now they've turned around and starting pushing this snake-oil, DRM'd "Hi-Res Lossless" nonsense.
The only reason to have "Hi-Res Lossless" is if you're going to do something besides listening with it... and you can't with Apple's streaming.
In the age of 4K,8K video streaming, free efficient lossless codecs and yearly $1000 phone upgrades why do we even bother to ask the question of lossy vs lossless audio for entertainment? It's not a matter of disc capacity anymore. You can have the best for "free".
If we do ABX test you will find out that people can't even make the difference between the original artist and a cover artist let alone lossy vs lossless. Should we just use cover artists at concerts?
The brain adapts quickly to lower quality be it visual, audio, olfaction or gustatory. Does it mean we should ingest the most we can tolerate because we get used to it so we can run on more efficient/cheap resources/content ?
No, you can't hear the difference, let's collectively move on, please.
Beyond comparing formats, even subjectively, it is important to consider how the public got used to compression artifacts.
I mean, the famous mp3 pre-echo was so common in early 90's that I think part of the listeners would prefer listening to it than to a cleaner sound. It is possible that mp3 influenced how music is composed, mastered and mixed.
That being said and adding the fact that people are willing to listen to music using cheap auricular phones in the noisy environment of their cars and recompressed using Bluetooth, I'd say that the 128kbps mp3 is still a very hard to beat format.
I came across this after listening to several 320kbps MP3 files and found that they sounded noticeably worse than 256kbps AAC versions. Support for AAC is widespread now and it should be preferred over MP3 [1].
[1] https://www.iis.fraunhofer.de/en/ff/amm/consumer-electronics...
> But what about when the codec is dealing not with a simple tone, but with music? One of the signals I put on Test CD 3 (track 25) simulates a musical signal by combining 43 discrete tones with frequencies spaced 500Hz apart.
Yes, but what about with music?
Interesting—have been tinkering in this area for decades now and always heard AAC was better than MP3. But until now have not seen how/why it was better. Thank you Stereophile.
Yes as several have written, the piece is from 2008 and it doesn't matter any more.
First, once LAME and VBR came about, I've never been able to tell the difference between my 192K MP3 and lossless files, even as a spring-chicken with expensive equipment. Been "good enough" for a very long time.
Second, since storage and bandwidth exploded I've used FLAC exclusively. Why not? But, have found 24/96+ files on the internet occasionally and first thing I downsample them to 16/48khz and do a listening test. I sure as hell can't hear the difference between those. I do leave the last extra 3.9khz... why not? Incredibly cheap and maybe the kids can hear it. Playable on car stereo and more compact, one third the size.
Finally, a big exception. Techies obsess about compression formats, but they don't matter as much as you think at the high-quality end. I've learned the source, i.e. master recording is more important. Example—rip "pristine" FLACs (or WAVs) directly from an iconic 80s CD. Do a listening test. Compare them with a modern remaster encoded with 192K Lame VBR MP3. The MP3 will sound a lot better and preserve the improved high end details. Yes, more noise but you'll struggle to hear it.
(Caveat—this is assuming we're not talking about a shitty 2010-era "loudness war" remaster but a quality-oriented remaster.)
Was mildly surprised by this after insisting on FLAC for almost two decades. A bit too early, in hindsight. Storage is so cheap now though, it again doesn't matter. FLAC it is, Opus from online sources.
Compressed music had a place, but in 2023, with 5G, unlimited data and streaming services, there's really no reason not to go lossless.
ALAC/FLAC files are pretty small, there's few downsides to going lossless. To be fair, there arent that many upsides either, but you at least skip one recompression step when sending the audio over BT.
For a while, HydrongenAudio, arguably the best sound/music related forum, did plenty of listening tests.
The last one sadly is from 2014: they tested Opus, AAC and Ogg Vorbis at 96 kbps against a classic MP3 128 kbps, and find out which codec produces the best sound quality.
https://listening-test.coresv.net/results.htm
https://listening-test.coresv.net/bytrack/index.htm
Notice that it is almost 10 years old, and that MP3 was encoded at 128kbps.
>compressed ilk do not offer sufficient audio quality for serious music listening.
Old old OLD argument here. Apart from -rare-, well-trained golden ears, very few humans can distinguish 192K or better, well-encoded MP3s without knowing -exactly- what to listen for. Or will need to 99% of the time.
As dozens of studies have shown over several decades. The rest is either marketing or self-deception.
I'm convinced that we can "hear" frequencies well above the reputed 20KHz limit of human hearing, as overtones, i.e. as tonal quality.
I certainly don't have golden ears; I'm no audiophile, and I'm getting on in years. 44KHz FLAC is easily good enough for me. But I tire of listening to MP3 music, after a few tens of minutes; it seems to lack the presence and immediacy that keeps me interested.
The intersection of not understanding digital audio and not understanding the neuroscience of hearing remains a place that never ceases to amaze me.
While acknowledging that I don't know whether I can tell the difference in every case or not, I would summarize my own preference for lossless audio in the following terms. Choosing lossy audio, my best case scenario is that I save space or bandwidth because I can't tell the difference; my worst case scenario is that I'm missing some element of the music, whether it is consciously noticeable, something I'm unaware of entirely, or perhaps something that I may only be experiencing on a somatic level that doesn't reach the level of conscious thought (I know that the possibility of this last option will be contested by some, and that's fair enough). Choosing lossless audio, my best case scenario is that I'm hearing the music in a higher fidelity, and increasing the amount I'm capable of appreciating; my worst case scenario is that I'm wasting some space or bandwidth for the reassurance. Basically, Pascal's Wager, but for audio.
Didn't use lame for mp3 so the conclusions are pointless in my opinion
This can be humbling http://abx.digitalfeed.net/list.html just try flac vs 128kbit.. I swear I used to be able to distinguish 320kbit, now I struggle with 96k :s Age and loud music...
2008.
We all know the Vorbis is supreme. Get out of here with your 15 year old DRM compression riddled subpar listening formats. OGG is all that matters. Without it… we wouldn’t have Spotify. <leaves before shoe is thrown>.
Wow, great tests! Now do a blind A/B with headphones on. :)
Anything is better than YouTube-- which seems to be the common format everyone is listening to these days. I would LOVE to be able to regularly listen to CD quality music...
There are people who can hear up to 22khz and beyond, so I've never understood why many of these arguments aren't easily resolvable on that basis alone.
I’m not entirely into the audiophile stuff, but from personal experience, you can tell the difference, last thing I tried was when I switched from Spotify to Apple Music, where the later has “lossless” option (I think even Spotify has that) but the difference was clear between the two for a streaming service, Apple one is just more clear and alive, I even opened the same song and kept switching back and forth between the apps just to make sure I’m not imagining stuff. Was it because the lossless on apple is better than the lossless on Spotify? Or something else? I don’t know.
Maybe someone can make an online quiz of a bunch of formats and accumulate statistics on how well people can actually tell or if it’s just a bunch of random.
For mp3, for me, 192kbps and higher is where it sounds pretty good, 128kbps sounds bad
Why do you guys tell others what they can or can't hear? We need to have someone do head surgery on you and implant a device to detect signals between your eardrum and the brain to see what's heard. Even then the brain probably has a lot to do with how sound is interpreted or processed.
Let's go over this one more time:
- Q: Can you hear the difference between CD-quality lossless audio and anything higher fidelity? A: No, no one even has the biological ability to. 44100khz, 16-bit audio can perfectly reproduce audio as far as we can physically tell. The only reason to store anything higher is for production or archiving (that is, for computers to listen to).
- Q: Can you hear the difference between 320kbps MP3 or the equivalent, and CD-quality lossless? A: Yes, this is _theoretically_ possible. However, many well-controlled listening tests have been performed on this subject that all say no, so it's much more likely that you can't, and the burden is on you to prove otherwise with an abundance of evidence.
e.g.: https://downloads.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP384... https://www.researchgate.net/publication/257068576_Subjectiv...
The listening test linked in the article leads nowhere, I would have liked to see their methodology.