Slashdot Mirror
Speculation On a Lossless iTunes Store
DrJenny writes "C|net UK has up an interesting blog post predicting that within 12 months Apple's iTunes Store will include a download center for lossless audio. This would be a massively positive move for people who spend thousands of dollars on hi-fi gear, but refuse to give money to stores that only offer compressed music — they could finally take advantage of legal digital downloads. The article goes into details on how Apple's home-grown ALAC lossless encoding relates to FLAC, DRM, and the iPod ecosystem."
Speculation On a Lossless iTunes Store
Lossless? I thought the iTunes store was a loss leader?
The theory of relativity doesn't work right in Arkansas.
Forget "lossless" when you've already lost so much of the original wave by mixing it down to 16-bit 44khz stereo in the first place. I'd rather have something that started out with a higher sampling rate/etc, but with good lossy compression to pull it down to something that doesn't require DVD-type storage for a single album.
I have seen the future, and it is inconvenient.
Hope this happens. After transcoding my CD collection to FLAC to arhive it, I now regularly batch re-encode to smaller and smaller bit rates using new releases of lossy encoders. AAC has gotten much better (esp AAC-HE) over the years to the point for a portable player, 48kbs is perfectly acceptable to my ears. With a 16GB iPod Touch, I could see buying music from the iTMS in some lossless format and transcoding to get my entire collection all on a small, flash memory player.
Linn Records offers downloads of 24-bit 96kHz songs. It would also be great to see DSD files available sometime. Those formats would really bring interest.
:)
It's good to see the possibility of lossless music nevertheless.
I'm tone deaf.
If you haven't made a developer cry, you've wasted a day.
From the blog:
"And now I have an inkling Apple will add lossless music downloads to the iTunes Store within the next 12 months."
Translation:
I have no fricken clue that this will ever happen, but because I think it'd be cool if it did, I'll go ahead and blog about it.
If someone says he and his monkey have nothing to hide, they almost certainly do.
Other than selection, which is arguably a non-issue these days, why would I bother downloading something as large as lossless audio when there's no real benefit to doing so? I could just as easily go to the store and pickup the original CD for only a small bit more than or, more than likely, the same price as the download. I get the physical media and it doesn't cost much more, this is a no brainer for me.
The ease of access argument is null, in my mind, because it has DRM and any ease is negated right there. When I spend the time to download FLAC from etree, dimeadozen, or where ever else, it's not a waste because the music is free, pretty much unavailable in any other format anywhere, and there's a huge selection of it.
I'm sure it will have limited success with those that are *that* excited about the delivery medium and are that obsessed with lossless format. For the rest of us that pretend to be audiophiles, we'll probably stick to our free FLAC files and/or purchased physical CDs.
Well, naturally they're not counting anything that happens before it hits the CD. CDs are the defacto benchmark. Yeah, it's not like seeing the band play live, but what recording is? This is all in the context of recordings.
import system.cool.Sig;
My kingdom for a mod point.
What about 24bits/sample, 96K samples/second?
That is also a very popular standard for audio, and is better than CD quality, by quite a bit.
What would be nice is a losslessly compresses 24/96 5.1/7.1 channel audio format to be their choice.
If I have nothing to hide, don't search me
Bravo to this -- enough with the 44.1Khz already!
Apple can cater to the portion of the market that has rejected AAC, while simultaneously ensuring lock-in by using their proprietary codec that isn't interoperative with other players.
It makes sense.
Win for Apple, and lossy everyone else, including customers. (Inless they have the wisdom to just say no and keep buying CDs. And iTunes store's popularity suggests lots of people don't.)
"Believe me!" -- Donald Trump
Most CDs have about 10-19 songs and range in price from $10-$15 (at least the mainstream ones). That works out to usually $0.99 a song. The last album I bought was Timbaland: Shock Value. 17 good songs for $12.
Isn't it amazing that 25 years after the release of the CD, we're excited to finally have a way to buy DRM free, lossless, digital music? If this happens, we'll be back inline with 1982 technology.
Nice idea, but Apple hasn't yet introduced the 1TB iPod Classic you'd need to hold all your uncompressed music...
I thought the whole point of having a music store online was so that they could make a tidy profit with minimal expenditure. Wouldn't that mean they've been lossless already?
Ask me about repetitive DNA
Or is it because of the definition of legal?
Sorry, Nyquist's theorem states that you can accurately represent frequencies up to 1/2 the sampling rate. Assuming you are a human and not a dog, you can not hear frequencies above 22khz. As for 16 bit, nobody uses all that dynamic range anyway. So 16bit/44.1khz is entirely good enough for listening.
Now 24/96 has its uses if you're mastering something, so that any errors introduced in the mixing process are below the quantization error in the final 16/44.1 product.
Give me Classic Slashdot or give me death!
I haven't RTFA'd, so don't know if this is in there, but Zunior.com is already offering FLAC downloads for $2 more than their mp3 downloads; it was definitely an incentive to buy from them, and I imagine some other, smaller mp3 stores are also offering lossless. Hopefully both Apple and eMusic will take the hint.
It wasn't me, it was the one-armed
Forget Apple... I updated my iPod's firmware to Rockbox (which natively offers several lossless formats, and a slew of other features) and haven't looked back.
I did this for 3 reasons... 1) iTunes stopped supporting Windows 2000. (Yes, I know it's old, but I don't have to deal with the stupid BS Microsoft has built into XP, like WGA). 2) The 1.2.1 Apple firmware for iPod Videos gave me trouble with a bunch of my MP3s--cutting off the song at the 75% marker and refusing to seek within the track. (Of course, the catch-22 is that I can't get a newer iPod firmware from Apple since they refuse to support W2K). 3) I never liked the way iTunes worked in the first place...
I don't hold out much hope that a lossless format sold thru iTunes will truly be lossless. After all, converting an LP to 16-bit 44.1KHz WAV is, by definition, lossy (but outside of the perceptions of 95+% of the people out there)... To add, part of the reason that iTunes even sells DRM-free music is because the record companies can say "if you want higher quality, buy the CD or, better yet, vinyl!" So, I doubt many record companies will be selling uber-high-quality lossless tracks through iTunes...
Windows 3.1x calc: 3.11 - 3.10 = 0.00
There is so much overmodulation and distortion in concert venues that one could argue that seeing the band play live is not like listening to the CD. Unless you're listening to an acoustic band in a small setting. You can't get much "more real" than that.
Can you be Even More Awesome?!
I'd buy the downloads only if they're DRM free, and lossless, because there's lots of artists that only have one or two songs that I like, and I would only want to buy those particular selections. One great example of such a one hit wonder is Melt with You, by Modern English. The rest of that album will make you pierce your eardrums. Or Chumbawumba's Tub-Thumping - the rest of the album serves as an incentive to push EMOs over the edge.
The cesspool just got a check and balance.
So they lock down these files with DRM. Then DVD-Jon (or someone else) comes up with a DRM-stripping program for the files.
Then people can re-encode the files to their format of choice. But by then, most consumers have said "fuck it" and decided to just download their format of choice directly from p2p or usenet because it's easier and simpler than paying Apple and still violating the DMCA just so the music they paid for will work on the audio player they own.
Oh wait, that's already the status quo... Never mind.
My thoughts exactly. Sometimes I think there's a bug in slash that prevents intelligent users from posting when I have mod points.
And now someone's gone and marked it flamebait... good grief... (I know, I know, meta-moderate.)
...make some noise; here's one place to start: http://flac.sourceforge.net/itunes.html
almost everyone else distributing lossless (except musicgiants) is using FLAC and/or WAV. it's supported by almost all s/w except itunes, hell you can even get wmp to play FLAC with some work.
re:TFA, lossless is not directly about quality, mp3 and aac both can be perceptually transparent for the most part, it's about (depending on your personality) perceived quality or format independence -- i.e. being able to transcode to the format you need without quality loss.
FLAC - Free Lossless Audio Codec
If you can afford thousands of dollars on equipment, why not just buy CD's and rip them with FLAC? Or better yet, have your butler rip them for you!
Said much better than I could have hoped to. I wish to subscribe to your newsletter.
More Twoson than Cupertino
Sure I might buy something in Apple's lossless format from iTunes, but
A - If I'm going to pay extra for DRM'd lossless, I better get the cheap lossy version for free (for my phone, wife's iPod, whatever) because paying them to compress a song for me is ridiculous,and
B - It will be a moot point if the player won't play all the FLAC I already have, because I won't own the player. It's why I don't own one now.
Operator, give me the number for 911!
Am I out of the loop? I was under the impression that most piracy was of the low quality mp3s that suck on any high end audio gear.
Lossless is a great idea and may open up a new market to the iTMS, but I can't image it's going to offset piracy. I'd think it will offset physical CD sales.
Help! I'm a slashdot refugee.
A witty use of tinyurl - mod parent up! +1 for non-obvious trolling ;)
Going on means going far
Going far means returning
the article claims that apple won't go with FLAC because we're against DRM. I don't think so; if we're to believe Steve then he's against it too. and there's nothing stopping apple from sticking FLAC in an mp4 container with fairplay, we can't prevent that anyway. aside from the principle of it, another reasone we're against it in FLAC is that DRM is doesn't belong in the codec layer, it's a layer on top.
apple's got nothing to fear from FLAC, it can actually be used to their advantage to get a leg up on the competition, since for lossless electronic distribution FLAC is becoming the de facto standard.
FLAC - Free Lossless Audio Codec
True, but consider CDs cut off is 20khz and 1 in 4 people hear as high as the 22khz you mentioned.
Really? I have a square wave, a sine wave, and a sawtooth wave, all at 22KHz. Now, you tell me how they'll be quantized such that all are accurately represented.
Either Nyquist is wrong, or you're misrepresenting his "theorem".
Do you have ESP?
You can't hear the difference between those three waves, unless your hearing greatly exceeds 22 kHz. The sawtooth and square waves are the addition of a 22 kHz sine wave and other, significantly higher frequency sine waves. Your speakers probably can't reproduce the higher frequency components, even if you could hear them.
Lossless audio is going to involve some large file sizes, and with that, comes increased costs--bandwidth ain't free, and storage/delivery of these files is not going to be cheap or easy. This all translates into fairly expensive downloads.
So for Apple to seriously consider this, they're going to have to figure out if there are enough audiophiles out there willing to pay that kind of money for downloads.
Personally, I kinda doubt it.
Gifts for Geeks - Stuff that really matters!
Yeah, I'm going to go ahead and say that when I set my audio less than 44 during recording, you can hear a very clear difference. It's not a golden ear thing, it's an obvious difference. Am I missing something?
Belief? Hope? Preference?The Existential Vortex
Selling 44.1khz/16bit/stereo audio files is the same business as selling downloadable DVDs. Only freak hobbyists will do it for the next few years until consumers have enough bandwidth, time, storage, motivation to indulge in such a hobby. Commercial big file size media business (excluding game binaries) is just a press release folks. All soundbytes, no usage.
And as for Nyquist... its a theorem, not a fact. Compare a 44.1/16bit file to a 96khz/24bit file on a studio grade sound system and you'll hear the difference. Nyquist didn't listen to music.
Your 22kHz square and sawtooth waves have higher frequency harmonic content. If you don't believe me, go work the fourier transform -- it's not actually that hard if you replace the sawtooth wave with a triangle wave. Regardless, what that means is if you took your 22kHz waves, and ideally low-pass filtered them at frequency f, 22kHz
If you don't believe me, you can (sortof) do the experiment yourself. Generate the waves at 1kHz and at 10kHz and play them back. With the 10kHz waves, they'll sound different, but they'll be much closer to each other than the 1kHz waves. You can do this on your 44kHz computer audio system. If you want something more convincing, do it in analog electronics and up the frequency to 22kHz. Finding speakers with good response beyond 20kHz is left as an exercise for the reader.
Nyquist's theorem states that a wave of frequency f must be sampled at the rate of at least 2f in order for information not to be lost. So, yes, a 44.1kHz sampling rate can accurately reproduce signals up to 22kHz without loss of information, and since that's all we can hear, we should be fine. Right?
Well, not entirely. You see, if the source material contains frequencies above 22.05kHz, they will end up "aliased" onto another part of the frequency spectrum. In short, the extra high-end becomes noise. Information is lost.
Here is the important part, in practical terms. In order to prevent aliasing, the source material must be low-passed to remove the unrepresentable high frequencies. Low-pass filters are not perfect; in order to toss out the frequencies we don't want, we end up attenuating some of the frequencies we do want. Thus it is not uncommon for high-frequency rolloff to begin in the mid-teens of kilohertz, even though we're aiming for 22kHz as the corner frequency.
This causes a real, human-audible difference in the finished product, and it is practically impossible to avoid.
Now, with a 96kHz sample rate, we aim to squash all frequencies above 48kHz, and our non-ideal low-pass filter starts to work in the 30kHz range. The imperfections in the low-pass filter are only apparent at frequencies humans can't hear. The finished audio ends up sounding like the source material, with no human-detectable loss in fidelity.
This is why 96kHz is a good idea.
Cretin - a powerful and flexible CD reencoder
learn to spell fucktard and maybe Ill visit that gay site you keep incessantly promoting....
Easy, a square wave(or any wave) can be represented (through fourier transforms) as a sum of sine waves of increasing frequency. If you have a 22khz square wave, what you really have is a 22khz sine wave, and a bunch of sine waves with frequencies greater than 22khz. Those higher harmonics cannot be accurately represented with a 44.1 khz sampling rate, but since you can't hear anything above 22khz anyway it doesn't matter.
Give me Classic Slashdot or give me death!
Lossless formats are a stupidly inefficient way of using up their bit quota. If we're allowed the same number of bits as a CD (or a lossless AAC) but instead we use it in some lossy format, then we can get much higher fidelity.
"Lossless" is a pointless criteria. The CD has already thrown away information (e.g. cut out frequencies above 22khz, cut out dynamic range to squash it to 16 bits, cut out anything more than 2 channels). It's silly to get enthusiastic about a "lossless" storage of this already-lossy data.
As much as it hertz, their loss results in your gain.
The theory of relativity doesn't work right in Arkansas.
Unless they are DRM'd, it doesn't particularly matter what format they're in -- you can transcode them to another lossless format, without loss! (Duh.)
Or you could transcode to mp3 and play it anywhere.
Don't thank God, thank a doctor!
You're missing 2 things. First, to get response to 22kHz or more, you need 44kHz or more sample rates -- remember, you can only represent frequencies out to half the sample rate. Second, there are lots of potential artifacts introduced by sampling. This includes aliasing artifacts both in the original sampling, and in any later sample rate conversions. Aliasing artifacts in the original sampling are (ideally) removed by an analog filter before the ADC; that filter may very well not actually be adjustable, so you get some sampling artifacts if you drop the sample rate. Second, if you use a sample rate that's not an even fraction of 44.1kHz, the ADC may very well upconvert it to a 44.1kHz stream before converting it into an analog signal for your speakers, introducing more artifacts. And if you drop all the way down to 22.05kHz, you've lost the 11kHz-22kHz frequency band, even assuming no artifacts, and that band is most certainly audible.
In short, do what you've already figured out -- if your system is designed for 44.1kHz, use it at 44.1kHz.
Now, of course Nyquist's theorem is correct. It's a theorem. Mathematically speaking it's unassailable.
But in practice, there's a caveat for real-world applications: by the same math, any frequencies over half your sampling rate that get into your source get converted into frequencies less than half your sampling rate. New, audible sounds that didn't exist in the source! And they sound absolutely awful.
So you need a filter that removes those sounds as aggressively as possible from your input with as little effect as possible on the audio you want to keep.
Here's the crux of the issue: 44kHz sampling only gives you a 10% frequency margin to go from "zero perceptible effect on the audio" to "completely blocks all audio". As it turns out this filter is pretty much impossible to build. Designers either compromise by allowing some rare aliasing noise (most audio equipment isn't designed to respond well above 20kHz anyway), or by starting the cut a little early (most people can't hear much above 16kHz anyway). As a general aside, the narrower and more accurate your filter is, the more delay it adds to your audio, so there's a latency issue, too.
It's much easier to build a filter that gradually cuts out audio starting somewhere above 20kHz and finishing completely (100dB or 150dB cut) by 40kHz for use with 96kHz sampling. And this is why 96kHz sampling is better: nothing to do with being able to hear over 20kHz, but merely engineering tradeoffs.
That said, you're right, most people can't tell the difference, or if they can it doesn't matter to them. And yes, you could probably record at 96kHz and use a (very computationally expensive) digital filter to downsample to 44kHz and produce something indistinguishable in mastering, but then I'm not actually a signal processing expert..
In electroacoustic music and classical music, 24bit sound does make a difference. I think it would be wise to stop at a standard that can at least produce sound that takes advantage of the compete rage of human hearing. Plus, people have easy access software that can do a lot of effects processing. Since re-mixing seems to only be getting more popular, I think the more the merrier, especially in sampling rate. Then you can do all sort of fun time expansion and pitch shifting without worrying about it sounding like crap.
Again, if Microsoft is about "Developers Developers Developers" then Apple's meat and drink is "Rumors Rumors Rumors".
Hail Eris, full of mischief...
E pluribus sanguinem
For serious?
If you've got thousands of dollars to toss around on audio equipment, you're seriously going to be stingy enough to illegally download music on the principle that you don't want to pay the $2-$3 more it costs to buy the physical CD?
I'm sorry, but that's got to be one of the lamest excuses for pirating music that I've ever heard.
To be fair, I actually can tell the difference between a 128kbps and a 192kbps MP3 when listening to certain pieces of music with a pair of decently nice headphones. 'Quieter' pieces, and most classical music don't do all that well under low-bitrate MP3 compression -- however, a 256kbps VBR MP3 (Amazon) is virtually indistinguishable from the original CD, whereas AAC (iTunes) is purportedly an even better codec.
Lossless audio is a waste of bandwidth, and frankly not worth the extra expense to the consumer.
However, if some music store wants to offer FLAC downloads for twice the price, I'm sure the audiophiles will be all over it, just like those $400 volume knobs.
(My prediction: The "They Might Be Giants" model of online distribution will become increasingly popular over the next few years, which will cut the music stores and record labels out entirely. As an added bonus, I'm sure a bunch of the bands will offer FLAC versions for a modest extra fee to appease their audiences.)
-- If you try to fail and succeed, which have you done? - Uli's moose
Why would people with high-end audio gear want to download digital music? I thought they all insisted on listening to wax drum recordings to achieve the best possible "natural sound".
I'm completely with you on the "nobody but the freakier people are going to notice", and they'll probably have gold-plated, gold-cable, etc. SACD players. Or, if they're really serious, they do away with the gold plating and have a goldsmith permanently goldsolder the wires right onto the board.
That said... the sampling frequency shouldn't be mixed with the signal frequency in the way you mention; e.g. 44.1KHz, divide by 2 (yay Nyquist), ~22KHz is the maximum frequency you can sample. ergo: 96KHz allows you to sample 48KHz signals and nobody can hear 48KHz anyway so what's the point.
Ah, true, but...
A 400Hz sine wave is now -also- sampled at the 96KHz level. Suddenly, that sine wave is looking twice as smooth.
Think of it like computer graphics. If you have a 320x240 15" display (12" by 9", non-widescreen 4:3), your pixel is going to be nearly 1mm on each side (12*25.4 / 320). A 1600x1200 display will have a pixel that is going to be much smaller, about 1/5th of a mm on each side (12*25.4 / 1600). Now you might not often find any reason to display a dot that is 1/5th of a millimeter at each side. However, if you were to display a large circle on the 320x240 display, it will be blocky. Do so on the 1600x1200 display, and it will appear to be much smoother.
Alternatively, find a piece of music that doesn't seem to do much over 22KHz, and band-limit it so that everything over 22KHz gets cut off anyway. Save this for later playback. Now actually downsample that to 22KHz. Now play back both files; see if you can tell the difference. Again, any high tones over 22KHz are gone anyway, so all you're hearing is the loss in fidelity of the lower-frequency ( 22KHz) signal.
I did, too. But then I switched back. Unfortunately, they didn't have support for realizing that the charger stopped sending juice to my ipod mini. So when I turned my car off, my ipod didn't automatically pause. This is a major feature for me with audiobooks.
As for 16 bit, nobody uses all that dynamic range anyway. So 16bit/44.1khz is entirely good enough for listening.
That's actually the biggest problem there. If they did use that whole range, CD audio would be flawless. Trouble is they usually compress it so it's all squeezed into a much smaller range.
At least so I understand it. My hearing sucks, so I'm quite happy with low bitrate mp3s:)
This is true, but doesn't account for aliasing: "Music, for instance, may contain high-frequency components that are inaudible to us. If we sample it with a frequency that is too low and reconstruct the music with a digital to analog converter, we may hear the low-frequency aliases of the undersampled high frequencies. Therefore, it is common practice to remove the high frequencies with a filter before the sampling is done." (from Wikipedia, which said it better than I was going to).
Very well put. It's one of the things that makes the delta-sigma modulation at very high sample rates used in eg SACD interesting. Of course, it would help if the data stream were easier to work with, which is why I think 24/96 or even 24/192 is superior overall.
The problem gets even more obnoxious if you care about the flatness and phase response of your filter. The one time I've done data acquisition work that cared about such things at 20kHz, we ended up using a 250kHz sample rate in order to give the Bessel filter room to operate. (We could have gotten away with marginally lower, but not enough lower to avoid buy the 1MS/s ADC system. We had 4 channels, so we ran at 250kHz.)
Does the source material contain those high frequencies? And secondly.. we can sample it at 98KHz and using the magic of DSP we can make a damned near ideal low-pass if we really needed one.
----
Go canucks, habs, and sens!
For apple. How about a mix of Apple and FLAC... We could call it AFLAC... that was bad.. it's considered polite to laugh at grandpa jokes.
Converting digital to analog regardless of recorded format will introduce noise unless you have a really good DAC carefully shielded from electrically noisy power supplies etc in your PC. Sure the DACs can be linear, have low distortion, but they better be in well shielded environment to gain any advantage over a 128 kbps mp3 file quality.
That is a good argument for why a studio should sample at a rate that accommodates the roll-off in their analog low-pass filters. However, once that is done you can use a can use a digital lo-pass filter / downsampler which can easily be designed to have very sharp cut-off rates. There is no reason at all for a consumer format to be more than 48kHz.
F-U-C-K-T-A-R-D. Now please visit my site as promised.
Now, go read the grandparent and find out that you're actually supporting my argument...
Do you have ESP?
I'd buy the downloads only if they're DRM free, and lossless
Honestly, I'd buy them if they are lossless and I can burn at least one audio CD. Because it's lossless, converting it to a CD and re-ripping it back as a DRM-less lossless audio file will not result in any artifacts. It works for me.
I think you're the only poster who's got it right in the specifics. The OP didn't exactly state Nyquist's theorem correctly... it is, if there is a BAND-LIMITED signal, I can exactly reproduce it by sampling at >= twice the band-limiting frequency.
Band limiting the signal ("anti-aliasing"... or your "analog filter before the ADC") is the key, because without this, there will be artifacts like you describe. And those ARE audible to our ears below 22 kHz.
Nyquist's theorem states that you can accurately represent frequencies up to 1/2 the sampling rate. That is 100% true. But in the real world, if you are sampling a digital recorder at 44Khz how do you ensure that NOTHING above 22Khz gets to the analog to digital converter? You need a strong analog filter but there are no filters that have an exactly square cut off Maybe let's say you have a 24db per octave filter. This mens you will have only attenuated the higher frequencies, not eliminated them. Same on playback. You need a theoretically perfect analog filter to playback. Such analog filers do not exist. The way they get around all this is to sample at 96 or 128Khz. If you do this then real-world analog filters can be used.
Most professional recording equipment now days records and mixes at 24bits, with final CD master downsampled to 16 bits. When used with proper dithering, the delivered 16bit, 44.1KHz CD will sound just as good as the master recording just before downsampling. You *do* know about dithering, right?
In Soviet Russia, articles before post read *you*!
You seem to have a good grasp, I was also under the impression that harmonics outside our audible range can still interfere with lower frequencies to produce a subtle but noticeable difference. Is there a grain of truth to this or is it just audiophile wishful thinking?
And just how good a representation of a 20 KHz or 21KHz wave does a 44KHz sample capture? Any beating effects? And how much music is a pure sine wave as opposed to superpositions of multiple waves that may have partial transient transitions of >22KHz?
Oh, so Nyquist's theorem doesn't assume an infinite precision sample and that dynamic range isn't a factor in a more accurate representation of small frequency variations? I haven't looked in detail at the math proofs for Nyquist's theorem but your interpretation doesn't match what I remember about SHM and wave superposition.
Well, one thing's true enough, very few popular recordings use the high end of the human hearing range now that Mariah and Witney are no longer in vogue. With an increasing amount of hearing loss in the general public, fewer customers of popular/contemporary music can hear much above 12KHz, at which point CD's are reasonably satisfactory.
Personally, I'm beginning to wonder if the real reason for dynamic range compression is so that customers aren't surprised by how crappy some manufactured idol bands and singers sound in person without heavy studio voice processing.
You might as well give it up. You're arguing with trailer trash, after all... You can't actually expect someone like that to have an education.
Personally, I'm beginning to wonder if the real reason for dynamic range compression is so that customers aren't surprised by how crappy some manufactured idol bands and singers sound in person without heavy studio voice processing.
What are you talking about? These manufactured idol bands sound exactly the same in person as they do with studio processing. After all, in their concerts, they're just lip-syncing to their recorded music.
Personally, I'm beginning to wonder if the real reason for dynamic range compression is so that customers aren't surprised by how crappy some manufactured idol bands and singers sound in person without heavy studio voice processing.
Watch out, you're about to start an argument with all the people who think that it's normal for good bands to make albums with only one good song.
I wish I could find the link, but there's reasonable evidence from blind listening tests that people, though they would not necessarily report any quality difference, were able to report things about the recording like "I can tell the cello is sitting in front of the viola" and other things that are very subtle and spatial. This of course depends on headphones and careful binaural recording, so on most end products it wouldn't make much of a difference.
In my line of work, most sound designers are recording all of their sound effects at 96K and 192K, a bit for the quality (guns and loud transient stuff sound totally bitchin), but also because if gives a great deal more latitude when you want to pitch down something-- you don't hear 30K overtones on an explosion, but it's nice to have them there when you pitch the explosion down 2 octaves, and your 30K overtones are at 7.5K and help keep the sound from sounding like it came over a phone.
I know a lot of you on this thread are arguing that 24 bit is worth it on an end product, but remember that the effective dynamic range at 24 bit is around 120 dB, which exceeds your threshold of pain by about 10 dB, so you're getting a ton of dynamic range that you're just going to use the volume knob on in the end to flatten out. Also, that implies you're listening in a silent room. Your average city apartment or townhouse has a noise floor around 40 db SPL at least, so you'd better have acoustical treatments or be on headphones that isolate that much (the more a set of cans isolate, though, the worse their spectral character tends to be, though.
Don't blame me, I voted for Baltar.
haha! a n00b troll!
The GP said:
Sorry, Nyquist's theorem states that you can accurately represent frequencies up to 1/2 the sampling rate.You said:
Really? I have a square wave, a sine wave, and a sawtooth wave, all at 22KHz. Now, you tell me how they'll be quantized such that all are accurately represented.You didn't say you had a frequency, you said you had a square wave. A square wave has a fundamental and an infinite train of odd harmonics; if you Nyquist-Shannon sample a square wave and then play it through a sinc()-reconstructing DAC, you'll only get the harmonics that fall below the Nyquist frequency. This applies to a square wave of any arbitrarily low fundamental, all non-sinusoidal waves that pass through ADC-DAC are rounded out because any steppiness preserved in the digital domain is smoothed out by the sinc function at the interpreting end.
I guess the detail the GP left out is that Nyquist treats all signals as CONTINUOUS ones.
Don't blame me, I voted for Baltar.
Except you are presuming that the human ear perceives a 20kHz sine wave, and a 20kHz sine wave plus a whole series of harmonics identically.
The problem with that notion is that the standard test for hearing perception is to play pure sine waves of varying frequencies and ask the listener if they can hear them. However over the millions of years of human evolution, it was not until the invention of the tuning fork in 1711 that any human ear had heared a pure sine wave. Up until that point it had evolved to distinguish multiple frequencies at once.
I am not aware of any scientific studies into whether the human ear is able to perceive the existence of harmonics in sound waves above what is considered the normal hearing limit. Surprising really because if it is, it would explain a lot when it comes to sound.
Argument over frequency manipulation and "true sound".
Oh wait...too late
> Those higher harmonics cannot be accurately represented with a 44.1 khz sampling rate, but since you can't hear anything above 22khz anyway it doesn't matter.
"The ear can't pick fundamental sounds at more than around 20 khz" != "the ear does a fourier transform and discard all harmonics above 22khz." The signal processing that a ear does to localize and identify sounds is a little more sophisticated.
I didn't do a double blind test, but even a seemingly small difference between a DAT recording at 44.1 and at 48 khz seems to make a slight difference in the sweetness of high end.
---- MISSING MISCELLANEOUS DATA SEGMENT --- [sigdash] trolololol
If someone can hear the difference between 256kbps AAC and a lossless codec without top dollar audio equipment, I would be very surprised.
Nothing against the poster, but I'd have modded it funny maybe. I mean I'm assuming the convenience and availability is the reason we down-graded our fidelity expectations. But it doesn't take an audio elitist to want their CD rips in, well CD quality.
When bandwidth was sparse and hard drives expensive this made sense. Then we've had to deal with the recording industry and their technophobia and related protectionism and lawsuits. But I really hope at the end of the day we aren't stuck with arbitrarily lower audio quality. Sound quality, like the related technology, should be increasing.
Quack, quack.
That won't work.
16/44.1 digital audio is already at the threshold of our perception. Ie, it's had to hear the difference between it and 24/96.
Lossy compression uses a perceptual model based on the way we perceive sound.
So, to create a lossy codec that is of a higher quality than 16/44.1, we need to know how we hear the difference between 16/44.1 and higher sample rate/bit depths. The available data bandwidth can then be concentrated on preserving what is most important to the human ear about that difference.
Since very few people can tell the difference, building a perceptual model of it is very hard.
As a musician and audio snob, I really prefer uncompressed audio, and when I got an itunes gift card and selected tunes to use it up, I saw they were 128k AAC IIRC, I was bummed. I wouldn't pay for that. I don't support the free music aka communist wave that is sweeping the industry, so if I am going to pay, I shouldn't get 2nd rate garbage for what I pay.
Where do you want to be, What are you doing to get there.
But that's not the point. I encode my CDs to FLAC, I can re-encode to any lossy or lossless codec I like without any degradation in quality. So it's perfect for archiving music. Or, indeed, buying downloads that I'm going to want to keep indefinitely. I see MP3s and other lossless codecs as something transient, an equivalent of cassette tapes - all right to listen to, but you wouldn't want to keep them forever.
You *do* know about dithering, right?
And dithering is essentially a hack to work around the limits of 44.1 kHz. In other words, the original response in regards to the benefits of higher sample rates still stands.
To quote someone a couple posts down with same same rant you had... And yes, you could probably record at 96kHz and use a (very computationally expensive) digital filter to downsample to 44kHz and produce something indistinguishable in mastering, but then I'm not actually a signal processing expert.. Expensive, exschmensive, you only have to do it once...
I don't think anyone has suggested that the original source should be sampled at 44kHz, only that 44kHz is appropriate for distribution. Why are so many people reading it the other way?
I became a FLAC fan in early 2001. Being able to put music on the hard drive without sacrificing quality was awesome. Back then, IIRC, the compression ratio was getting significantly better with each (pre-)release, and the coolness of transcoding from lossless to lossless became apparent. Thanks for all your work on making this great format available to all.
You do know that most studios record on 16bit 48khz equipment, right? That 4khz doesn't make much of a difference. In fact, most studio masters are slap-dash affairs. Bad mikes, bad recording equipment, inadequate space, etc. All that crap puts all but the very best masters far below what CD Audio is capable of. In this real-world context, there is no point at all to formats like SACD and DVD-Audio. What people actually WANT is pretty clear. People want CDs with a 5.1 Dolby Digital mix, or an equivalent surround system. The studios have been highly-resistant to introducing such a format because it costs 10X as much money to master surround sound recordings as it does plain stereo.
I really don't understand all this audiophile crap. Most of the sources are so lousy there is little point in trying to optimize your equipment. This is sharply contrasted with videophiles, because the movie studios actually bother to master their DVDs (and before that, LaserDiscs) properly. The same is generally true of Blu-Ray and HD-DVD. pull it down to something that doesn't require DVD-type storage for a single album. This makes no sense. Lossy compression can introduce nasty effects and can kill your range. Even the best psychoacoustic models (like LAME) still have serious problems with certain tracks. An uncompressed album fits in 650 MB, far less than a DVD (9 GB). Using FLAC or a similar codec would get that down to about 350 MB, less than many of the video downloads on iTunes.
It's still a lossy format which strips out some of the audio detail. I'm no gold-connector-magnetically-balanced-shielded-cable audiophile, but I do appreciate being able to listen to the entire depth of a piece of music (especially classical).
Perhaps a better way of putting it would be 'the human ear cannot distinguish between 320kbps MP3 and FLAC if listened to on iPod headphones', which is fair enough. There's no need to include everything if all I'm going to do is listen to it on the bus. Which leads to my original point - MP3 is lossy. AAC which is my format of choice is better quality for the space and bitrate, but is still lossy. FLAC isn't, which means I could have my lossless FLAC copy on my desktop where there's easy storage space, then have iTunes automatically create reduced quality versions for carrying around on my iPod. Compression from lossless source is always better than compression from an already compressed copy.
Not to mention that the iTunes store *isn't* 320kbps. 128kbps for the normal content, 256kbps for iTunes Plus.
How many people can read hex if only you and dead people can read hex?
check out www.mindawn.com they have been flac and ogg since the beginning, anyone can get their material on there and you can preview any song up to 3 times in full in any browser.
For me it's also a bonus that they are already in a digital file format (which I what I want because that's what's compatible with my my iPod, my phone, PC/Mac and home audio system). If it came on CD I'd have to rip them then chuck the CD in a box - rather than just click "Buy" and start listening to the track/album right away. I also find it handy that it's automatically downloaded and organized by iTunes, which stores them just the way I like (and it's configurable) - it also works just fine with the 'iTunes Music' folder mounted via NFS of a Linux system where I store/backup/serve my data from.
Personally, I have no problems with the 512 Kbps non-DRM'd tracks at present. Quality wise, what pisses me off is the lousy quality of the videos on the iTunes Music Store (that and the so-limited-it's-worthless selection of TV content on it here in the UK). That, and the videos are DRM'd of course. I would note that if the Apple TV didn't have a cripplingly restrictive defective-by-design implementation with regard to the way it accesses content remotely OR you if could at burn the video's to DVD then in practice the DRM would not particularly bother me. If they only allow 5 computers to be authorized able to play your DRM'd files at any give time, there is clearly no need for the Apple TV to insist that you can only send content to it from one of those computers at a given time - but it's the same BS with the iPod.
Although (going off topic a bit here) it's fair to say would rather have a much wider range of music on the iTMS, it's also fair to note that can't usually get what I'm after unless I go to a very large store either (e.g. like a large Virgin Megastore, as was) or to a second hand/specialist store - and I don't even think my tastes are that eclectic.
The RAW equivalent for audio would nice, but lossless would be what it would take for me (and everyone I know) to buy online.
If any of you remember cassettes, low end MP3s are about equal (IMHO).
I haven't bought / downloaded any music because of this factor - it's just not good enough when I can purchase the CD and deal with it from there.
AAC is pretty damn good, but no, I can tell the difference for the most part and well, really, come on, get real - they already SELL it lossless, it's not like you're twisting knobs to transfer it to the hard drive.
If anyone can get the majority of the Corporate Music above the line brain dead to listen, it'll be Jobs, and Team Apple, both of them.
~hylas
I definitely disagree about the dynamic rate in certain cases. (I'm an audio engineer, DSP programmer, and electro-acoustic musician) The problem with dynamic range is that you have the same number of bits to represent 0 to -6 Db (mono) that you do for -6 to -inf Db. Once you get down to the softest sounds, you often don't have many bits left at all to represent the sound; you only get the full range for the loudest of sounds, and the distribution of bits is linear, while the distribution of loudness is logarithmic. For most music, it's not a problem, but it does cause problems for orchestral works which can have a huge range of dynamics. For instance, in Messaien's Éclairs sur l'au-delà there's a bit with two triangle players on opposite sides of the stage playing rolls as soft as possible, and that never sounds right in recording. Also, there's plenty of electroacoustic pieces that benefit from the increased dynamic range.
If you record at 16 bit but allow 12 Db of headroom just in case of really nasty spikes (which can definitely happen with an orchestra), then you are now effectively recording with 16384 possible values instead of 65536, meaning only 84 Db of resolution with which to record. For most listeners, though 16bit/44.1khz is fine, and it absolutely destroys vinyl in terms of fidelity, aliasing, etc. be damned.
A good chunk of the stuff on iTunes is basically DRM free - it's tagged, but that's it. All EMI, gradually a lot of independents, too. And if you can burn a CD (as you can with DRMed stuff in iTunes), and what you are burning to CD is lossless, well, I don't see the problem here.
Argh.
D/A converters have not used analog filters since the end of the last century.
They filter in the digital domain, and oversample before the filtering.
The '256X oversampling' on your CD player really does mean a D/A converter sample rate of 12.28 MHZ!
Filtering is much easier with such wide bandwith.
Whether 44.1KHz or 96KHz, both get upsampled.
WTF is a mod point?
Iraq billions
None of this matters if I turn it up to 11, though.
When they came for the communists, I said "He's next door. Take him away. Goddam commies."
Close, but not quite exactly right. The theorem proves that for a given sampling rate, it is possible to precisely reproduce any signal that includes only frequencies <= 1/2 the sampling rate. However, if you sample a signal which contains frequencies higher than 1/2 the sampling rate, you do not get data sufficient to precisely reproduce just the frequencies up to 1/2 the sampling rate. Instead you get (sometimes nasty) artifacts of the the higher frequencies showing up in the reproduced signal.
Assuming you are a human and not a dog [lib.unb.ca], you can not hear frequencies above 22khz.Literally true, but incomplete. Humans cannot hear pure tones above 22kHz; many of us cannot hardly get over 16kHz. But we can hear the harmonic effects from the mixing of frequencies up to about 30kHz. I believe that this may not have been understood when the CD standard was created; I know it has been proven by double-blind studies. If you're familiar with what a second-harmonic beat frequency looks like, then imagine what adding a 24kHz signal to a 12kHz signal does, and tell me you don't think the difference would be audible.
As for 16 bit, nobody uses all that dynamic range anyway.
Just because it's not used for pop/rock doesn't mean it's not useful for anybody; there are types of music that have tremendous dynamic range and 16-bits is really not enough.
I filter out all the frequencies above 22kHz. Now I sample. at 44Khz. Now they're all represented the same. The square wave and sawtooth wave differed only in higher-than-Nyquist-frequency (>22kHz) harmonics anyway.
OK, suppose I master at 96Khz. Then I apply an ideal 22Khz filter (easy enough once it's digital), and resample down to 44Khz. No aliasing. Well, I guess there's Gibbs artifacts, but they're unavoidable anyway.
Dude. Lossless should mean that if you buy a song and your computer explodes later, you should be able to re-download all the songs you bought. Free.
That would essentially be DRM free. :)
The cesspool just got a check and balance.
Ahh, so you don't know about dithering, because your statement makes no sense.
Dithering is principally related to sample depth (eg 16 bit), not sample rate (44.1 kHz). Dithering is not a hack; it's quite ingenious and well researched. It has nothing in particular to do with 44.1 kHz CDs and everything to do with fundamental issues of digital sampling. Dithering (and it's fancier cousin, noise shaping) minimize quantization distortion. You want this. You need this. Done properly, the "noise" that it adds is invisible, while the distortion it eliminates is quite tangible.
No matter how fast or how deep you sample, you will still want dithering.
I've had audiophiles* just snub their noses at mathematical proof and regrettably inform me that I do not have "the golden ear." I wonder if there have ever been any research on whether self proclaimed audiophiles REALLY have magical hearing.
I always laugh at these idiots who have no grasp of the technology. When sampeling an analog waveform, the more the better. When copying a digital signal bit for bit is perfict. Someone care to redefine lossless? Why sample 24 bits/sample when the digital source is 16 bits? Why not do a lossless copy bit for bit of the original 16 bit source?
"Bit rate = 44,100 samples/sec × 16 bit/sample × 2 channels = 1,378.125 kbit/s (10.09 MByte per minute)"
http://en.wikipedia.org/wiki/Red_Book_(audio_CD_standard)
Just send me a FLAC of the original CD and I'm happy.
The truth shall set you free!
Yes. Absolutely. However, the frequency represented between any two samples (a sample is a measure of the amplitude of the wave at that point, btw, not the frequency) is NOT constant. In fact, with a minimum of 3 instruments (guitar, drums, human voice), 6 strings per quitar plus reverbs and echoes and everything else that makes music what it is, the frequency varies inside the 1/44000ths of a second between two samples. So whilst, according to Nyquist's, any CONSTANT frequency can be sampled perfectly accurately at 2f, a series of frequencies overlaid onto one another to form a constantly, quasi-randomly changing frequency cannot.
QED
I wish I had mod points right now to mod down a bunch of these totally
ignorant descriptions of how bit-depth/sampling rate affect sound quality
Anyone that says "we don't need the dynamic range of 24 bits" has no fking clue
how digital audio works. 24 bits is better because it allows for a smoother waveform,
such that it more closely resembles the infinite voltage range of analog. 16 bits is
is a relatively "jaggy" waveform. Most people perceive it as "harsh" even if they don't
understand why.
I think I'll go tell some doctors how to use a scalpel since I sorta know what the word
scalpel means.
Well, not entirely. You see, if the source material contains frequencies above 22.05kHz, they will end up "aliased" onto another part of the frequency spectrum. In short, the extra high-end becomes noise. Information is lost.
Aliasing is noticable on most CD players I have tested. It is measureable and is most common in the range of 14KHZ to 22KHZ. To test it yourself, get a DDD created sweep test CD and play it. As as the sweep goes up, the diffrence frequencies come down and is plainly heard as decending tones in the sweep. It is quite visiable in a scope display. I know this from years in the consumer electronics industry. I used to sweep systems. I could not use the test CD for a pro sound studio setup due to the artifacts in playback unless the studio was for vocal range only from 100HZ to 12 KHZ. For recording brass instruments, I had to use a digital sweep source for certification other than a CD as a CD player was not up to par for pro studio work.
This is one of the best test CD's produced. It is all digital except the live samples.
http://www.amazon.com/Denon-Audio-Technical-Various-Artists/dp/B0000034ME
List all tracks to get to track 65, the sweep tone.
You can listen to the sweep signal online. The artifacts are fun in the online compression. This is one of the few test signal CD's that start at 5 HZ.
Just watch the limited bandwidth and compression eat this test signal alive online. Remember, this is a digitaly created source signal. Any dropouts and roll-off and artifacts is in the compression and playback. Check out track 65 for a 5 HZ to 22KHZ digitaly created and ruller flat in amplitude. This will acid test your equipment.
A FLAC of this CD is lossless and has the same response as the original CD.
The truth shall set you free!
... why we are still having this argument? I would have thought this would be an open and shut case. People complain about low sample rate, other people show those people that it's scientifically impossible for a person to hear the difference. Are there just different people each time, who have never heard of Nyquist's theorem? Or do they feel that science isn't representing their experience? This isn't a troll; I'd really like to know.
You know, there is a difference between trolling and pointing out the flaws in your reasoning. Just saying.
I think Nyquist's theorem is utter BS. Let's just think about this for a second, shall we?
44.1KHz gives us literally 44,100 opportunities per second to measure where the wave is at. With me?
Humans can hear up to 22KHz. That's two samples per cycle. Think that's giving you an accurate depiction in any way, shape or form?
OK, maybe you can't hear 22KHz. At 11KHz you have 4 samples per freq. cycle. Possibility of a nice sine wave and not much else.
Anyone, even your grandfather, can hear 5.5KHz. 8 samples per cycle. Better than 4 but hardly accurate. Why don't you try reducing a graph of, say, a stock price over even a day to 8 samples, and then translate it back, without losing information.
Even a reference tone of 1KHz gives you only 44 and a bit samples per cycle. That is totally inadequate to describe a complex sound in all its detail, whatever Nyquist says.
And pretty much anyone with any experience of live vs. recorded music knows this to be absolutely true. I've never mistakenly confused a CD for a real-life sound, and I doubt you have either.
So, don't pull out Nyquist as if it's the ultimate end to all these arguments. A little common sense will tell you that information is being lost, a LOT of information, and there's ample room for improvement - SACD was a nice idea, but useless if you didn't record with it in the first place. I'd like to see some of the ideas from SACD, though, filter through to the recording community at large - it's a genuine leap forward if it can be implemented for everyday recording.
They are there, no one but the dog can hear them but they are there and the mic picks them up not well but it picks up 33Khz.
Not all mic's. Most vocal microphones do not go that high. Check the spec.
The old industry band vocal mic, the Sure SM58 response is here;
http://www.shure.com/stellent/groups/public/@gms_gmi_web_us/documents/web_resource/site_img_us_rc_sm58_large.gif
Sure SM57;
http://cachepe.zzounds.com/media/sm57-0e448d5589fdc8d6658bd863b801f637.pdf
The newer Sure vocal mics are here;
Sure PG58 http://cachepe.samedaymusic.com/media/pg58-d7a8418e0d8d830ff025d91f4eef8a58.pdf
Sure PG48 http://www.shure.com/stellent/groups/public/@gms_gmi_web_us/documents/web_resource/site_img_us_rc_PG48_large.gif
Even ditching a dynamic for the better condensor mic gives this response;
http://www.americanmusical.com/manuals/shure/shusm86_userguide.pdf
Some instermentation microphones may extend into the ultrasonic range, but most are flat through 20HZ-20KHZ with a rapid roll-off above 20KHZ.
The truth shall set you free!
This applies only to me, of course, as I was the only test subject however here are my findings:
I can stastically distinguish up to 256kbit/s MP3 vs FLAC.
At 320kbps my rate is around 65%, which is not sufficently higher than 50% to declare that it was distinguished (over my 20 tests).
The following equipment was used:
Sennheiser HD650
Benchmark DAC
Fed using Emu 1616 from computer.
Using my ZD5's from ZaphAudio (www.zaphaudio.com) which I built, I had less accuracy due to noise level in room.
Tests were done double blind using Foobar's ABX test application. Test tracks were Mel Torme - Sleigh Ride (Jazzy Christmas with Telarc), Herbert von Karajan - Beethoven's 9th (Mvmt 4) and Rachael Yamagata - Worn Me Down.
At least for me, FLAC is not by any means an absolute neccisary. The portability options for conversion to other formats is a huge factor looking forward however. I am sure that those with Ipod earbuds would have less resolution capacity, however my ears are not by any means extraordinary.
The Nyquist Theorem is mathematically provable.
A house divided against itself cannot stand.
Why would you sample at 44.1KHz? Sample at e.g. 4x and down sample digitally to 44.1KHz (BTW, the down-sampling should be done after mixing/mixer and the mixer should have bigger resolution than 16 bits).
In the playback a damn good oversampling digital filter is norm for a CD player nowadays.
Why would you sample at 44.1KHz? Sample at e.g. 4x and down sample digitally to 44.1KHz (BTW, the down-sampling should be done after mixing/mixer and the mixer should have bigger resolution than 16 bits).
In the playback a damn good oversampling digital filter is norm for a CD player nowadays.
What about if you want to convert it to another format (maybe your phone only plays .WMP format, and you've got an iPod that doesn't - or maybe some new fancy format comes out next year)? If you capture it in anything other than lossless, the quality is going to get worse every time you convert.
Are you capable of having this discussion without sounding like a complete asshat?
One reason people love FLAC so much is that it copes well with inter-track gaps, i.e. the gaps between songs on a CD. Some CDs have no gap between songs, they just run into each other ("gapless"). You can play back MP3s encoded with LAME as gapless, but it's a hack. Using EAC, you can make perfect bit-exact FLAC rips that play back exactly as the original CD did, including gapless parts.
I'm betting iTunes won't support it. It took them years to get gapless playback working on the iPod, and they never back-ported it or offered fixes for people who had already bought DRMed tracks. When you factor in the cost of buying a whole album of DRM-free tracks vs. buying the CD from a discount on-line store, it doesn't look good. In fact, often the CD can be cheaper, especially used on Amazon.
Oh, and lets not forget that the sound output from iTunes isn't brilliant anyway. At least on Windows, it goes through the Windows mixer which re-samples everything to 48KHz. Anyone serious uses WinAMP or Foobar with Kernel Streaming/ASIO to bypass that.
No serious Hi-Fi buff will bother with this, they will stick to CDs or pirate FLAC rips.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
This is true (at least for me - I already can't hear the difference between CD and 192kb AAC). However, the idea here is probably to let the user archive the music, and encode it in the format he or she wants. 320kb AAC sounds awesome, but if your player only plays MP3, you've already got one transcoding action which will take away from the quality of your music.
I pretty much assumed the 160 GB iPod Classic was going to be the end of the line for that... Maybe a few people have more than that amount of music, but not terribly many. And the screen, compared to the touch model, isn't great for viewing video... Add uncompressed music and all of a sudden, 160 GB is likely to seem cramped.
.aac's altogether, just to kind of force the point.
So, yes, on the one hand, I'm sure audiophiles at Apple thought it'd be a great idea. But on the other, I'm also quite certain that the marketing department also thought "what a great way to continue the growth of the sales curve"
Next, they'll stop offering compressed
... a download center for lossless audio. This would be a massively positive move for people who spend thousands of dollars on hi-fi gear, but refuse to give money to stores that only offer compressed music
Whoever will be going to set up such a thing should keep in mind that people who spend thousands of dollars (make this dozens of thousands, if you like) for their audio equipment are not just picky when it comes to the format of the tracks they are offered.
They are equally picky in terms of recording/encoding quality of the tracks, to start with.
They won't be willing to deal with any kind of DRM, watermarks and other annoying side effects.
And last but not least, they are especially demanding when it comes to content. Real content, top quality content. For many of the audiophiles this will be Classical music, featuring the world's top orchestras, conductors and voices, including a tagging scheme that finally makes sense in operas and symphonies, etc. etc.
This could be achieved, but then we're not in Kansas... err... mainstream music business anymore, as we know it today.
Otherwise most people in the target group will just laugh all the way to their favourite downtown CD-and-vinyl stores.
That wasn't me, above. I know what dithering is, and it's great, but all it does is swap one kind of noise (periodic stuff, like sampling and quantization error) for another (non-periodic noise) which we as humans don't care about quite as much. It's not a magic bullet. It will not magically make 16 bits able to hold the information of 24. It just makes those 16 bits sound better.
Cretin - a powerful and flexible CD reencoder
The AVERAGE human ear cannot distinguish the difference between a *perfectly encoded* mp3 at 320kbps.
However, not all MP3 codecs are perfect, and not all ears are average.
Just as some people have a sense of taste so powerful that they can tell you every ingredient in a sauce, every spice in a stew, some people have an ear so precise that they can (easily) notice the muffled treble and dropped notes.
Not surprisingly, these people are often the ones buying the $80,000 stereo systems.
I see MP3s and other lossless codecs as something transient, an equivalent of cassette tapes - all right to listen to, but you wouldn't want to keep them forever
I beg to differ. I have tapes that are older than many slashdot readers, dating back to at least 1970. Many of them I've replaced the housing on, as the rollers jammed or the lubrication sheet wore out. Some have been "eaten" leaving flutter in spots, some have been spliced leaving a skip, but some sound so good that when I've sampled them to CD people can't believe that they were originally cassette.
With analog, your gear matters greatly. My car is an '02 with a disk changer and cassette. A home burned CD of a cassette actually sounds better than the cassette it was sampled from in the car!
There is a tiny audible difference between most factory stamped CDs and their cassette counterparts, but on some analog-recorded ones the cassette CD sounds better than the factory CD due to bad digital remixing from the studios.
Cassettes recorded and played on good equipment are vastly superior to MP3s at any bitrate and come pretty close to CD.
-mcgrew
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
The Nyquist limit says nothing about the shape of the waveform, and if you think about it is just common sense - at the nyquist limit there is less than one sample per crest, so if you didn't filter out all frequencies above the limit you would get very audible nastiness.
A 15kHz tone has three samples per wavecrest. With only three samples there's no possible way to differentiate a square wave from a sine wave from a sawtooth wave. With digitally sampled sound, the higher the frequency the more aliasing.
And although your ear can't hear the tones above CD's nyquist limit of 22kHz, those tones color the frequencies you can hear.
-mcgrew
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
There is no reason why anyone would need more that 640KB.
For want of a nail the shoe was lost
For want of a shoe the mod point was lost
For want of a mod point the troll was lost
For want of a troll the battle was lost
For want of a battle troll the kingdom was lost
Therefore, mod points enable trolling. Brought to you by the slashdot logic system.
-mcgrew
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
Nyquist's theorem states that you can accurately represent frequencies up to 1/2 the sampling rate
The nyquist theoorum says nothing about accuracy. It says that it is impossible to represent frequencies above 1/2 the sampling rate. The higher the frequency the more aliasing. At 44k samples per second a 15 kHz tone has three samples per wavecrest, so you can't tell what the shape of the wave is at all.
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
That limit is only in determining the frequency. However, beat two waves at double the frequency but same volume and then do the same two frequencies but at half volume. Now don't they sound different?
The allocation of amplitude is only feasible when you have a few samples going up and down, where frequency only needs two up, two down. The more accurate you need the amplitude, the more samples. Then there's quantisation errors (so 24bits help), errors in isochronity (so sampling at 3x frequency, you are necessarily more accurate, even if it's a larger % of the interval). All these add up to the Nyquist half frequency limit being far less than the whole story.
Errors in sample frequency, errors in quantisation, noise. The Nyquist theory is about FREQUENCY. But there's a lot of difference in a piccolo played fortissimo and pianissimo (spelling). A 10kHz square wave sounds different to a 10kHz sine but to get that difference, you need four more harmonics (three if you're lucky), which means you're right at the limit with a 10kHz sound. And that's not all that high.
It's a little like the theory that bees can't fly. It's just that in Real Life, there are things that aerodynamic theory doesn't have to deal with in theory but in real life, they can make a difference.
You don't understand it at all. :(
Nyquist states that there is *exact* reproduction of all frequencies 1/2 Fs.
You only need two and a bit samples to reconstruct a sine and it's phase, and all waveforms can be thought of as a sum of sinewaves of different frequencies and phases.
So, as any sine below 1/2 Fs is reconstructed exactly, so is any possible waveform with harmonics 1/2 Fs.
Arrrgh. The slashcode has removed all the 'less than' signs from my post.
The first and last times I write '1/2 Fs' should have a 'less than' symbol before them.
The poster above is correct. The OP's information is almost 20 years out of date, and he is moderated upward. Oversampling has been used in even consumer players for over 15 years. This poster corrects him, and the moderators take points from him. This is where Digg is much better than Slashdot. A smart user like this guy would have been moderated upward. Instead here where normal users are not allowed to contribute to the moderation, we have morons that don't understand the topic making stupid decisions. Come-on Slashdot, make the moderation useful. Open it up to everyone.
Hey, I didn't pick the guy's name. He calls himself "Trailer Trash"; go look for yourself. What am I supposed to think, that someone who admits he's trailer trash would be highly educated? I don't think so.
You don't get it.
It might help to remember that any waveform can be reconstructed by the summing of a series of sine waves at different amplitudes and frequencies.
As long as the sines that make up the waveform are less than half Fs, exact reconstruction is possible.
There are no frequencies between the samples. Even thinking about that shows a deep misunderstanding of the subject.
Remember all frequencies above 1/2 Fs are filtered off before A/D conversion.
I agree that iTunes should support FLAC or at least let you rip from it to Apple Lossless.
The workaround I use is to convert the FLAC to WMA lossless, which iTunes can then import.
You could also burn it to a CD and then rip it. A two step process either way.
I'm a 2000 man.
"Even a reference tone of 1KHz gives you only 44 and a bit samples per cycle. That is totally inadequate to describe a complex sound in all its detail, whatever Nyquist says."
Ah, but it is adequate. Any 'complexity' not recorded contains frequencies above half the sample rate. Any sine wave can be described as long as you have more than two samples.
"Anyone, even your grandfather, can hear 5.5KHz. 8 samples per cycle. Better than 4 but hardly accurate. Why don't you try reducing a graph of, say, a stock price over even a day to 8 samples, and then translate it back, without losing information."
If the stock price is low pass filtered in the same way as digital audio before conversion, this is not only possible, but trivial to do correctly.
1 bit SACD works with exactly the same theory as multibit PCM audio. They are no different in principal.
So, couldn't you use a 192khz ADAC and an ideal digital filter to get it down to 48khz?
My 96khz DVD-Audio disks have much better sounding cymbals then their CD equivilents. I'm not sure if it's due to the higher output sampling rate, higher bit density, or 5.1; but I will say that I prefer higher sampling rates when available.
No, I will not work for your startup
No commercial product has used an analog 24db per octave filter before an ADC since some time in the last century.
They use a gentle RC filter and oversample the ADC.
Most modern ADC designs run at a fixed rate in the MHz range, and decimate the output to 44.1Khz or 96Khz or whatever. Almost all are delta-sigma converters.
The *analog* filter slope and ringing artifacts on the input remain the same regardless of output sampling frequency.
Same applies to D/A converters. This is common in all audio electronics nowadays, as it's cheaper to do fixed high sample rate low bit depth silicon than to do good steep analog filters.
Quit whining about sampling rates and gold cables and such. Just go see the Melvins or Motorhead live and your hearing will come down to the level of normal people who just enjoy music.
Pooty tweet
MP3 was a far more entrenched audio format than FLAC when Apple came out with the iPod, but that didn't stop them from making AAC the default format for iTunes and the iPod. And for people who already have iPods or even just use iTunes, what's the incentive to switch to FLAC? Just rip to Apple Lossless and call it good.
"Waveforms" are merely a convention we teach children. The way our ears work is a lot closer to a Fourier analysis machine that doesn't even detect phase. There are no square or sawtooth hair cells.
[citation needed]
The evidence for hypersonic hearing is extremely sketchy so far, and even the pioneers have only found unconscious effects. If it works it will merely become the MSG of music, added mechanically to make the album seem better than it actually is.
"Waveforms" are merely a convention we teach children
You've never seen an oscilloscope? There is indeed a marked audible difference between a sine wave and a sawtooth wave and a square wave. This is exactly what a guitar "fuzzbox" does, turns the guitar's (imperfect) sine wave into a square or sawtooth wave. Mine has a switch that lets you use square or sawtooth (not labeled as such but that's what it does) and a foot switch that turns it off (passthrough) or on.
Fourier analysis machine that doesn't even detect phase.
Yet a phase shifter (sold at music stores everywhere) gives a very marked difference to the sound.
There are no square or sawtooth hair cells.
You don't understand what sound is at all! The cells detect pressure and nothing more. There is no "square pressure". Rather, a sine wave crest is a gradual increase in pressure, tapering off, then gradually dropping. A square wave crest is a sudden all at once full force which is held steady for a marked period of time (depending on its frequency, or rather vice versa) then dropping instantly to zero.
An oscilloscope graphs this pressure. The squiggly lines on an oscilloscope show pressure; the higher above the zero mark the line reaches the greater the pressure (and volume), and the lower below the zero mark the same, only in the opposite direction.
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
What you see on an oscilloscope better describes what happens in an amplifier, but that doesn't make it more real than what you see on a spectrum analyzer--and the latter better describes what happens in your head.
In the frequency domain, a square wave is a fundamental sine wave plus odd harmonics. This is not merely a mathematical abstraction, it's what our ears are physically equipped to detect. A hair cell that resonates at 15 kHz just responds to any periodic pressure changes that arrive about 67 ms apart, it cannot distinguish waveforms. Our brains could work out waveforms by combining that with other hair cells that resonate at 45 kHz, 75 kHz, and so on, but we don't have any.
In the frequency domain, a square wave is a fundamental sine wave plus odd harmonics
One frequency is the same as another to a signal generator, provided the frequency is within its range. With the ear's hair cells I don't know, but I've heard no other explanation why I've heard LPs that could convince me that there was a live human with a musical instrument in the room, while I've never heard a CD that even coes close.
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
I find that surprising, I didn't think anyone expected to get better than 20 kHz out of a vinyl LP beyond the first couple of plays. If I had to guess, I'd blame the 16-bit linear quantization used on a CD (there aren't many distinct amplitudes near zero to represent quiet sounds) but nobody seems to come out and demonstrate such big differences in A-B tests.