Slashdot Mirror
Apple in Talks to Improve Sound Quality of Music Downloads
Barence writes "Apple and music labels are reportedly in discussions to raise the audio quality of of the songs they sell to 24-bit. The move could see digital downloads that surpass CD quality, which is recorded at 16 bits at a sample rate of 44.1kHz. It would also provide Apple and the music labels with an opportunity to 'upgrade' people's music collections, raising extra revenue in the process. The big question is whether anyone would even notice the difference between 16-bit and 24-bit files on a portable player, especially with the low-quality earbuds supplied by Apple and other manufacturers. Labels such as Linn Records already sell 'studio master' versions of albums in 24-bit FLAC format, but these are targeted at high-end audio buffs with equipment of a high enough caliber to accentuate the improvement in quality."
Looks like I'll have to bootleg my music collection all over again.
Godaddy is a scam and a ripoff.
How can I appreciate the awesome response of my Monster Cable speaker wires if I'm not playing 24-bit FLAC audio files over them?
I've abandoned my search for truth; now I'm just looking for some useful delusions.
Audiophiles listen to stereos. The rest of us listen to music. :)
A correctly mastered 16-bit file wouldn't have any audible difference compared to the 24-bit file anyway, unless we're talking measurable differences instead of differences you can actually hear. I'd rather see an increase in the samplerate, but preferably both.
c++;
Apple will of course rebrand FLAC as Apple-FLAC, or AFLAC for short.
the bit depth is interesting, but the largest improvement would come from simply not using lossy compression. one hopes that TFA glossed over this and that nobody is seriously considering 24-bit MP3's.
with an alpha channel
Perhaps is music wasn't overly compressed (talking about dynamic range, here) they wouldn't need so many more bits of resolution for the -3 dB they're mastering audio at these days.
More Twoson than Cupertino
Over the decades I've read several studies testing peoples opinions of different bitrates and compression schemes. The typical response is people can just barely tell that there is a difference between bitrates, but they are unable to accurately pick the HIGHER bitrate one. In other words, even when they can tell there is a difference, they're still not sure what one is the original...just that they sound "different".
I don't even want to get started on "audiophiles". They're institutionalized hatred of the sound of live music sickens me...they claim to want the best quality possible, but won't suffer through anything that hasn't been run through an unintentional distortion or dynamic range limiting filter.
Not really, although possibly, depending on the recordings. The difference between 24-bit and 16-bit audio is the dynamic range, with 24-bit having a much wider range between the quietest possible sound and the loudest possible sound. This is something that can definitely be heard, even on lower end equipment.
16-bit audio has a 100dB dynamic range and if properly dithered from 24-bit to 16-bit almost no one will notice the difference. To claim otherwise is to fly in the face of ABX tests which back this up.
A quick note about dynamic range, which is what the bit depth affects.
Maximum dynamic range that human hearing can discern: 140dB average
Maximum practical dynamic range of CD: 90dB
Maximum practical dynamic range of 24-bit audio: around 140dB
Dynamic range required for full range live music playback, according to Ampex: 118dB average
Maximum practical dynamic range of high quality studio analog tape: 80dB
Maximum practical dynamic range of studio analog tape in the '60s: ~70dB
So, if you have a piece of music recorded, mixed, mastered and released in pure 24-bit depth, you *may* hear a difference under ideal conditions (excellent production, good equipment, *quiet* listening room, etc...) Note that there have been double-blind listening tests of SACD, and listeners were unable to hear a difference between the CD version.
All those old Beatles and Rolling Stones albums? Keep the best CD version you have, more bits aren't going to make a difference.
My Other Computer Is A Data General Nova III.
Sure they'll be 24-bit, but they'll also have the dynamic range compressed to shit.
Unless that's the actual selling point, getting copies of the songs before they've passed through the hands of the mastering engineer whose job it is to destroy the difference between the quietest and loudest parts of songs, or worse yet causing horrible clipping.
What I'd like to see (or rather hear), is that we can have access to the individual tracks of each song, so that we can remix stuff. Kind of like the open-source of audio.
If Pandora's box is destined to be opened, *I* want to be the one to open it.
From an audiphile forum:
20% of the money will buy you 90% of the sound...another 30% of the money will buy you another 5% of the sound...you can't buy the remaining 5% of the sound because nobody can agree about what it is.
Not just audiophiles. I can't hear worth crap. Never could. I could hardly care less for the difference between a scratchy record and a CD, much less what color my cable is, gold or green or fuscia. What I do care about though, is being able to format shift my music. My archive is in FLAC, which I transcode to a lossy format for general use. When something more palatable comes along, I'll be able to transcode to that instead of having to repurchase everything -- that assuming I could even find half of it, which is very unlikely. And unlike if my collection were solely in a lossy format, I won't have to endure the progressive distortions of transcoding from one lossy format to another, the cumulative effect of which would eventually drive even me nuts.
128kbit/s at 24-bit! Now excuse me while I crank it to 11.
My stereo(yes, two channel!) is worth several thousands of carefully-planned dollars. I think it could be put alongside systems worth $20k, and hold its own. (The speakers at present are the weakest link, and they still sound much better than yours. :-)
That said, it's a practical system. I've got enough background in electronics and acoustics (and psychology!) to know better than to buy a huge amount of the insane junk that's out there. Amplifiers that go into oscillation with the wrong cables? No thanks! Vacuum tubes? The guitar amp is downstairs, thanks very much. Cable elevators? Um...no. Just no.
So here's my defense of 24-bit 48kHz recordings: Breathing space.
Nothing to do (specifically) with dynamic headroom or the like, but when producing, mixing, and mastering data recorded as 16-bit 44kHz, there is very little you can do without inadvertently affecting the audio signal. In other words, it's harder to get it right when you're operating right at the threshold of hearing.
If studios did everything in 24-bit/96kHz and actually avoided clipping through the whole chain, then a final mixdown to 16b/44Khz (i.e. a CD) would sound gorgeous - perfect sound to the extent of human hearing. However, mixing is often done poorly, and as hot as possible for better sales, and the result is that the poor CD suffers the abuse caused by the engineers.
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
CD have more then enough dynamic range, it's just that it is hardly ever used.
Analogies don't equal equalities, they are merely somewhat analogous.
>>>Today's music, however, is so compressed (as in audio-compression, not data-compression) in the quest to "make it louder" that it doesn't even get close to reaching the possible dynamic range of 16-bit, which effectively makes an upgrade to 24-bit completely worthless.
>>>
Ding-ding-ding-ding-ding. We have a winner!
As for quality I used to care, but not anymore. As long as the MP3s I download.... er, I mean purchase sound as good as the FM Radio where I originally heard them, that's good enough. ----- If an artist releases a Greatest Hits CD I'll buy that, but mainly to "support" the singer with his commission, not because of quality.
BTW Super Audio CD and DVD-audio failed because nobody cared about quality. I expect these 24 bit things to fail too. If Apple really cares about quality, they should start selling Lossless versions of their songs.
Information wants to be expensive AND wants to be free. So you have Value vs. Cheap distribution fighting each other.
If you don't think there is real high end equipment it just means you've never looked/listened. I'm not taking $1000 speaker cables or other such snake oil, I'm talking high end speakers and so on.
Speakers in particular have a wide range because they are almost always the worst component of a system. An amp that has THD in the fractions of a percent may be hooked in to a speaker that has THD in the 5-10% range when played at a high volume.
There can be a pretty big difference between normal and good equipment. There's also a pretty big monetary difference so it isn't worth it for everyone, but if you like good sound, maybe it is. It also isn't something magical that you have to have faith exists, it is stuff you can measure. Flatter frequency response, lower THD, lower noise, better dispersion, etc, etc.
Now, does that mean 24-bit is useful? Eh, I dunno. In theory possibly. You get 96dB of dynamic range out of 16-bit audio. You can extend that through dithering, but at the cost of raising the noise floor. Human hearing is more in the 120dB range. 0dB SPL (20 micropascals) is chosen as 0 becuse it is roughly the threshold of human hearing. Some people can hear a little below that, many cannot hear that low because of hearing damage/loss. 120dB SPL is about the level where you start to feel immediate pain and thus going past it is not recommended.
So to fully cover the human range of hearing you'd need 20-bits, but then more can be useful because of course if you are trying to represent low level sounds with just 1 or 2 bits, they are going to have rather bad quantization artifacts. Again dither can deal with this, in trade for higher noise levels, but just going 24-bit solves it.
As a practical matter though, it is of questionable usefulness. For recording it is quite useful because it allows for headroom. You want to be able to have plenty of digital headroom (to prevent clipping), but still capture all the detail. However when you mix everything and normalize it down, that's not so important. It also takes some fairly high quality equipment to start getting 100dB or more of actual effective SNR and dynamic range out of a system, not to mention a rather quiet room. You can hear sounds below the room's noise level, but only maybe 10-15dB below.
I've played with it quite a bit since audio production is a hobby and I really can't form an opinion. I can set up tests where I can hear the difference, but I can set up tests where I can't.
Over all I think it would be nice to move to 24-bit since space is rapidly becoming a total non-issue and it just avoids it ever being a problem. Kinda like moving past 8-bits per channel for video. However I don't think it is a big issue and it isn't something I'll tell people they gotta have. "CD quality" has endured precisely because it is "good enough" for most things. Maybe not perfect, but you don't really notice any problems in normal use and that's what matters.
A year or two back I decided to actually test where I hit transparency on MP3s encoded with LAME as it then was. I thought I'd get transparency at somewhere between 192kbs and 256kbs. I didn't, I got it at roughly 160-170kbs depending on the song. (Too many cymbals does fuck that up but up the bitrate enough and at maybe 210-225kbs cymbals and cornets go transparent for me, too. Maybe I've got cloth ears but since I'm encoding my music for me I don't give the slightest hint of a fuck - also I doubt it, I think the desire to prove how great you are is driving a lot of audiophiles to convince themselves they can hear more than they can.)
A good friend of mine -- a better musician than me by a long way and I'm not actually that bad -- hits transparency at about 150-160kbs on modern encoders, though to be fair he uses OGG by default and I tend to hit transparency down around there on OGG too.
I'd love to see more people who claim they *need* lossless to listen properly do an actual, full double-blind on a range of tpyes of music. I've no issue believing other people hit transparency higher than I do, but frankly I don't believe anyone who says that 320kbs MP3 isn't good enough.
Disclaimer 1: I have FLAC rips of all my CDs except a few which I ripped with iTunes and haven't swapped from ALAC yet. This is partly to have full quality archives of my CD collection, and partly because I'm well aware of the haemorrhaging of quality you get by reencoding compressed files.
Disclaimer 2: Both I and my friend did these double blinds through headphones. Nice quality headphones (Sennheiser over-the-ears, can't remember which model) but headphones nonetheless. The results through a big speaker stack would probably be different and I'd expect to hit transparency a bit higher, at maybe 256kbs again. But I might be wrong and it might be lower (or, of course, it might be higher).
Disclaimer 3: Not really a disclaimer, just that when I record at home I tend to record in 48kHz and 24 bit. That's mainly because my computer is aging. Give me more RAM and I'll happily sit there and record at 96kHz and 32 bit. I'll then downsample it to 16 bit and 44.1kHz because I really don't see the point of doing anything else given I'm compressing everything so that maybe -50dB is the lowest volume my music hits and normally it's between -30dB and -0.1dB...
Disclaimer 4: I'm very loathe to buy anything from online music stores because they're only offering compressed formats, so I've automatically taken a quality hit. But when there's no physical release I buy them anyway because in all reality I can't pretend to tell the difference between a 320kbs MP3 and a CD and nor can I tell the difference between a 192kbs AAC and a CD. But if I ever have to re-encode -- like if I end up back somewhere running Linux and I can't put on MP3 support unlikely as that now seems -- then I know I'll lose some quality. Which may or may not be audible, of course....
This is the stupidest thing I've heard in a long time.
16 bits per channel gives you a whopping 96dB or so of dynamic range.
All popular and contemporary music has crap compressed out of the dynamic range so you'd be lucky if you could get a discernable 20dB of range. Classical music needs a lot more, but not 96dB. Maybe 60 or 70dB.
You would need as quiet a listening environment as an empty concert hall, and a very high powered amp turned up loud, to even hear as much dynamic range as is represented by 16 bits.
And they think adding more even dynamic range than that is a good idea?!
If they wanted to make a difference to sound quality, they should increase the sample rate to 48kHz, or hey go why not crazy and go to 96kHz. It will still not sound any different to the average person, but at least the difference can actually be detected and measured with the right equipment.
This is just a prelude to the new Apple iEars implantable neural audio interface (with full DRM and iAd support) that they're going to sell you so you can fully appreciate this exciting new bandwidth. Then there will be the iEars TruSeven 7.1 channel version, which involves drilling another six holes in your head so you can actually experience BluRay Movies the way that God intended.
The Genius Bar guys can get you set up with an appointment at your nearest AOSC (Apple Outpatient Surgery Center).
G.
The biggest determinant of sound quality is the recording and mastering. Most music coming out these days is mixed for portable players, and is made to be heard through a torrent of street noise. It doesn't matter how good your bitrate or bit depth are, if the track has a 10-15dB dynamic range, and clips throughout the song.
A great recording, on the other hand, demands an adequate sound system. On my reasonably expensive system (MSB DAC, Aragon preamp, B&K amp, Klipsch speakers), I have done blind A/B testing, and was able to tell the difference between a 320kbit MP3 and WAV. On the other hand, using OGG, I was reduced to only being able to tell the difference up to 256kbit at best (depending on source material). My mom, who's a conservatory-trained musician, was able to pick out 256 kbit OGG from WAV 100% of the time (total of 10 tracks), and 320kbit OGG vs. WAV on about half of them. My guess is that a professional musician might do even better.
We don't really need 24-bit recordings. We need the producers to use 24-bit in the studio, and then a nice 16-bit final output with dithering, and we have all the dynamic range we really need.
16-bit gives you about 90 dB. That's enough to go from "barely audible in a quiet room" to "starting to make your ears hurt". It's enough dynamic range, really.
But look up the "loudness wars" and find that much music being sold these days doesn't even use all that dynamic range. They compress the daylights out of the music to make it "louder".
So, I'm sort of interested in the 24-bit standard, if and only if it implies that the music will be produced with some actual dynamic range. If Rush makes a new album, they can release the CD with the dynamic range compressed away to nothing; and they can release the 24-bit mastered with some actual dynamic range.
Will this actually happen? Who knows. But I'm cautiously optimistic. This will give the studios the chance to release two completely different mixes, the mass-market one that "has to be loud" and the one marketed at audiophiles which "has to be clean". I don't spend $2000 on a power cord for my stereo, but I do appreciate a clean mix, so I hope this does work out.
steveha
lf(1): it's like ls(1) but sorts filenames by extension, tersely
Most mastering engineers will "cheat" and pump the dynamics on quieter parts of a long symphonic composition on account of the medium, and then they'll accentuate dynamic changes by tucking the faders leading into sforzandos (and that's not even getting into mic placement for room versus ensemble emphasis, cheating spot mics into the mix, varying polar patterns and EQ during the performance...). You really don't get a realistic dynamic of what was originally recorded from a finished off-the-rack CD -- the recording engineers know they're being graded for how dynamic it sounds and how the performance is translated to the medium and not necessarily how accurate it sounds. They really want to have it set up in such a way that the person at home never has to touch the volume knob.
Thus, I'm tempted to say that if something sounds like shit on an MP3, it probably wasn't mastered very well, because if the sound is suffering it means that the mastering engineer is letting the recording drift outside the listener model and are letting the pianissimos get too pianissimo because they can, even if it means dragging the program through the dithers, which is probably why the MP3 is suffering, it's wasting a lot of signal space encoding the mastered dither. There are some engineers that are really pedantic and are really touchy about ever using mixing, and they want to force people at home to have to strain to hear the quiet bits, but these guys are all nutty audiophiles who go home to soundproofed living rooms equipped with Klipsches.
Don't blame me, I voted for Baltar.
You will still experience better sound quality with a 16-bit CD than with 24-bit FLAC files.
This is because CDs are recorded and played back with lasers, meaning they are essentially an analog medium, because light is analog.
FLAC files, on the other end, are made of bits - sharp little bastions of absolute certainty, having a value of either or one, with nothing in between.
So, in essence, while CDs are recorded in analog bits, whereas FLAC files are recorded with digital, electronic bits. This means that CDs will sound better, because each bit is closer to the original recording; and, remember, that the whole is even more than the sum of it's bits!
PS: Make sure you orient your Monster Cable wires in the appropriate direction to maximize electron flow! There should be an arrow on the insulation pointing toward the speakers.
Do daemons dream of electric sleep()?