Slashdot Mirror
Master Engineer: Apple's "Mastered For iTunes" No Better Than AAC-Encoded Music
New submitter Stowie101 writes "British master engineer Ian Shepherd is ripping Apple's Mastered for iTunes service, saying it is pure marketing hype and isn't different than a standard AAC file in iTunes. Shepherd compared three digital music files, including a Red Hot Chili Peppers song downloaded in the Mastered for iTunes format with a CD version of the same song, and said there were no differences. Apple or someone else needs to step it up here and offer some true 'CD quality downloads.'"
You want CD quality downloads? Yeah, magic keyword "FLAC".
Piracy: giving you for free what the market won't since the first bestiality video was filmed.
While I agree that its all bunk, I would be interested in knowing if the two files where bit for bit the same or just sound the same to the listener?
"Have you ever thought about just turning off the TV, sitting down with your kids, and hitting them?"
Summary is incorrect. Article says that there was a significant difference between the Mastered for iTunes and CD version, while there was no difference between Mastered for iTunes and a standard AAC track.
To test with Red Hot Chili Peppers is rather pointless, I would think - they're one of the most compressed bands there is, probably not using more than the top 4-5 bits out of 16. So yes, it's going to be fairly similar no matter what the format, unless you can get ahold of the sources to the original masters.
Here is the actual relevant part of the article:
I am not an Apple fan, and, actually, dislike the company. However, I wonder if they are truly making claims that are not true or if their claims are simply carefully worded to convey, well, nothing. Apple seems to big and way too self-important to risk the scandal of an outright lie. It reminds me of how they handled the antennae problem with the iPhone. It would be interesting to hear Apple's response but my guess is that they will simply not respond and their fans will be fine with that.
http://www.busyweather.com/
From the /. summary:
Shepherd compared three digital music files, including a Red Hot Chili Peppers song downloaded in the Mastered for iTunes format with a CD version of the same song, and said there were no differences.
That'd be a good thing if there were no differences between the CD version and the audio versions. However, what the article actually says is
After his comparison of the three digital music files, Shepherd says there was a sonic difference between the Mastered for iTunes waveform and the CD waveform. He says the Mastered for iTunes and AAC-encoded files didn't reveal any differences, adding that this proves to him Apple's Mastered for iTunes isn't any different than a standard AAC file from Apple's iTunes store.
In other words, the Mastered for iTunes version is basically identical to the standard AAC version, and both are different from the CD version.
GLaDOS for President 2016! "Well here we are again. It's always such a pleasure." -- GLaDOS, 2011
Those wonderful color screens people could put on their TV's to impreove the picture -- you can't get more out of something than you put into it. If the lossy music process has lost data you can't put it back (but you can always convince the gullible that you can!)
Now, buy my Slashdot Post Converter, which placed on your screen turns each of my posts into a fantastic media experience! Zowie!
A feeling of having made the same mistake before: Deja Foobar
Summary says: "Shepherd compared three digital music files, including a Red Hot Chili Peppers song downloaded in the Mastered for iTunes format with a CD version of the same song, and said there were no differences." Emphasis Mine. If there were no difference, then this new format sounds great; what's the problem. Oh right, slashdot.
TFA says: "Shepherd says there was a sonic difference between the Mastered for iTunes waveform and the CD waveform."
Ugh.
Put an end to over compressed music and squeaky sub base encoding with WAVE files! Uncompressed WAV files are super, you can make them as big as you like bits and Hz wise! Need 8.1 channels? NO sweat ... that's 4 stereo and 1 mono! Works great with other uncompressed formats too! Compression is for tiny old handheld gadget weenies, to summarize ... :0)
The purpose of existence is to make money.
Or get the LP version of the album :)
It's The Golden Rule: "He who has the gold makes the rules."
This "mastered for itunes" stuff is pointless crap as long as we are still fighting the Loudness War.
The Red Hot Chili Peppers are a particularly bad test case because all of their albums have massive loudness-compression. And the same guy responsible for that travesty has started to do the mastering on recent Metallica albums so their stuff is going to be all suck too.
When information is power, privacy is freedom.
first:
"Mastered for iTunes format with a CD version of the same song, and said there were no differences. "
then
" Apple or someone else needs to step it up here and offer some true 'CD quality downloads."
Isn't no different then the CD version CD quality?
The Kruger Dunning explains most post on
The summary link just goes to a (slow loading) blog post, the actual article being discussed is at:
http://productionadvice.co.uk/mastered-for-itunes-cd-comparison/
And more specifically, the 11 minute youtube video:
http://www.youtube.com/watch?v=kGlQs9xM_zI
I can't agree with Shephard's methodology here. Why is he using RHCP to guide these comparisons? These are notoriously poorly mixed & mastered tracks - repeat offenders in the Loudness Wars debates that consume the audio engineering industry. Start with a record that is mastered to audiophile standards (say, RATM's debut) and then let's see if there's a difference. A piece of shit is going to sound like a piece of shit no matter what. Especially when the original (CD) master has digital clipping...
Further: "No differences" between CD and downloaded quality is a GOOD THING: The "Mastered for iTunse" AAC format is a compressed version of the audio data found on a CD (both are 16 bit, 44.1khz) so no discernible difference would mean that the encoding is transparent, does not result in a loss of dynamic range, and that the auditory masking effect does not leave artifacts.
While you are probably right about the RHCP (as well as any other CD released lately) about the gain compression, there is no reason why you still cant do quantitative analysis to see differences between the two different compression schemes when they are decompressed. It would just be a matter of comparing the bytes of decompressed files to the original, never compressed (ie from the CD) data. Compute the variance of the errors in the two different schemes referenced to the CD and the check to see if they are statistically significant. There you go, you have just tested to see if the compression schemes are different.
That being said, you would probably have a better indicator of errors if you did use a source that was not heavily gain compressed before data compression, but that is another debate.
I can't speak for the RHCP tracks, but I downloaded a dozen or so tracks I already have on CD and exist as both FLAC and LAME MP3s on my computer on the day of the announcement to see what the difference is. I could immediately tell a difference with the Master for iTunes tracks, better or worse, I'm not sure yet. They are easy to pick out in A/B testing, the most glaring difference is in the mid-bass area 80-120hz is noticeably boosted in the rock tracks I downloaded.
Stupid Cheap Guitars
It's interesting that this sudden focus on compressed music as opposed to uncompressed (iTunes Plus) has cropped up so soon after Steve's demise. IIRC, Steve was a music nut and was always pushing for DRM-free, higher fidelity digital downloads through iTunes. My foil-hat says that this might be an attempt to sell shitty quality music at a higher price. However, it could also ease network burden when streaming audio on the go. That said, one should still have access to high quality, uncompressed music for when you want to pump up the volume on your home system.
HOW different though? I am willing to bet that people tested on very good stereo equipment can't tell the difference between a 320k MP3 and redbook audio. With their ears - not with waveform equipment.
And I mean a real blind test, not just playing both and having them claim that one sounds better.
Is CD quality really the holy grail of audio quality? I thought DVD Audio with up to 24-bit bit depth and 192kps sampling rate was supposed to the the best in audio quality - far beyond the human ear's ability to hear.
Or is CD Quality "good enough", even for audio engineers?
"Mastered for iTunes" is just a set of guidelines that ensure that the resulting AAC file is the highest quality possible when encoded directly from a 24-bit master. It's higher quality then most FLACs because they are usually 16-bit, whereas AAC is essentially 24-bit when the source material is 24-bit. In essence, compressing 24-bit audio to 256kbps AAC sounds better then going to 16-bit uncompressed audio.
If you're going to go FLAC, at least make sure that you're getting 24-bit.
No, I will not work for your startup
Or, for one of the albums, you can get the MFSL version.
The gold plated CDs and ultra-heavy LPs may be a gimmick, but they do know how to mix masters.
I know many friends who have used higher compression on their FLAC files and, with my gear, I can clearly hear the artifacts. I realize most people won't but I've got mostly high end stuff, and I always burn in both my audio and network cables before using them and mark them with directional arrows (only with pvc-free tape and audio-grade markers) so that the don't get installed backwards after they've been burned in.
I'm amazed at how many people can't seem to grasp the fine points of lossless compression for audio work. I find most non-audiophiles expect that lossless means that what you put in exactly matches what you put out. I can tell you first hand, though, that when you spend as much money on gear as I have, you recognize that perfection comes from not just the bits, but the purity in which the bits are delivered. They may be the same ones and zeros, but a discerning ear can always tell the difference in the various lossless formats when listening to the color and soundstage of the reproduced performance.
Is it just my observation, or are there way too many stupid people in the world?
I want Vinyl quality.
This completely misrepresents what the 'Mastered for iTunes' represents.
If give the producer the tools and options to create CD quality files.
If a producer is putting a mastered for iTunes stamp ion the song that hasn't been improved beyond the most filmiest technicality, then it's on the producer.
There are a lot of issues regarding Apple products, and how Apple runs it's business. Lets not try to make some up, m'kay?
The Kruger Dunning explains most post on
"Apple recommends to mastering engineers not to compress a track. Owsinski says there’s a feature called Sound Check in iTunes that lets users hear all of their music files at the same level. He says a highly compressed track will actually sound less impactful because of the Sound Check technology. Moving beyond some of the technical aspects of mastering he adds that compressing is becoming less relevant because file storage is becoming less of an issue for consumers. This is why he theorizes that Apple is pushing its AAC Plus lossless high-resolution format (The existence of AAC Plus Lossless could not be verified)." .wav file and turning it into an mp4/AAC/Ogg/pick your poison.
The second falls within Applke's domain as it addresses the type of file and it's size as delivered to consumers.
The first has nothing to do with Apple. I'd love to see an Apple employee try and tell Bob Ludwig how he should be using his racks of high-end tube compressors and limiters. That would be a short conversation.
Audio compression during the mixing and mastering process has nothing to do with compressing a digital file to make it smaller. The first is a way of limiting dynamic range which is useful for several reasons such as creating a perceived loudness, making audio levels more consistent, or changing the character or transients through attack and release time settings. The other is taking a
Compared to what? People who buy what could be described as music files on a plastic disc?
Isn't that the point? This is confusing: " no difference from the CD " ... "someone needs to offer CD quality downloads..." .
The point for "Mastered for iTunes" is not to make it different from the CD, it's to make the compressed, lossy AAC file as close to the CD as possible. It sounds like they've done that.
That being said, you would probably have a better indicator of errors if you did use a source that was not heavily gain compressed before data compression, but that is another debate.
No, that was actually my point. If you use the CD as a master, and the CD is heavily compressed (as is the case here), there will be far less difference between different compressed versions. You're not going to get anything that sounds better than the master you use, cause the bits that are gone are gone. And in the case of RHCP CDs, that's unfortunately most of the bits.
tl;dr: GIGO
whatever the world champion of compression is nowadays
My vote is for Nickelback's latest. I have never seen a waveform so deprived of any information - it's on or off, with absolutely no headroom.
It depends on the equipment and room. As important though, is the program material. Take pretty much any chart release right now and it's irrelevant as those tracks are pretty much squashed to death to begin with. Something like Joni Mitchell's Blue, or Miles Davis, or well engineered orchestral music ... yes, there is a difference and it's noticeable. Then again, it's my job to notice (audio engineer)
What good is an optimized (if not lossess) format when played through and iPod with a digital to analog converter that costs 50 cents? My $40 Sansa Clip plays FLAC and has a better quality DAC than a $300 iPod. Why is Apple even bothering?
Here's what the article actually says: If you have a sound engineer who creates a recording with material that is badly distorted in the first place, then whatever Apple tries to do with "Mastered for iTunes" is not going to help, and the AAC encoded material sounds the same as AAC encoded material converted from a CD.
According to the article, the recording itself is not clipped, but it sounds as if clipping has happened at some time earlier in the production. Garbage-in, garbage-out principle.
First problem:
Subtracting one waveform from another to look at the difference, doesn't prove there are audible difference in the case of AAC vs CD.
It just proves that one file is using perceptual encoding, which we already know.
Perceptual encoding changes the waveform, but that does not prove that the difference is audible when in the original file being masked by louder material. To prove that you would need Double blind listening tests.
So that point is a total failure.
Second problem.
"Mastered for iTunes" produces 24 bit files, this offers more dynamic range than CD. But in order to see a benefit, your source material also needs more than 16 bits of dynamic range. You are lucky if RHCP has 10 bits of dynamic range.
So again, nothing proved.
Do you not get CD-quality sound when you download a song in Apple Lossless format?
I have a CD (I think it was either "Machinehead" or "Bad Company" that has a very noticeable hiss on some systems.
It reminds me of when I played tapes in crappy decks, and makes me wonder if the CD was recorded from a cheaper analog system. On tapes it probably would have been *less* noticeable as many decks actually had noise-reduction stuff build in.
This is a store-bought CD, not a burn.
From the article: Shepherd compared three digital music files, including a Red Hot Chili Peppers song downloaded in the Mastered for iTunes format with a CD version of the same song, and said there were no differences. Apple or someone else needs to step it up here and offer some true 'CD quality downloads.' So let me get this straight, this guy compared the Mastered for iTunes format with the CD version and found no differences. Then the headline says someone needs to "step it up here and offer some true 'CD quality downloads". So given the first statement, it seems to be that Apple is in fact offering 'CD quality downloads'. Am I missing something here?
Apple's "Mastered for iTunes" is a set of guidelines about how to turn a master recording into an iTunes-optimized digital file. The author of TFA, however, is talking about taking a CD track and making a compressed version that's as close as possible to the CD track. A CD track is NOT a master file. (We don't want a track that's merely a CD representation - we've heard plenty on
So of course if you make an iTunes track from a CD track via the "Mastered for iTunes" process, you'll get a 256 kbps VBR AAC that's identical to ripping a CD track to a 256 kbps VBR AAC. However, if you follow Apple's recommendations, quoted here:
To take best advantage of our latest encoders send us the highest resolution master file possible, appropriate to the medium and the project.
An ideal master will have 24-bit 96kHz resolution. These files contain more detail from which our encoders can create more accurate encodes. However, any resolution above 16-bit 44.1kHz, including sample rates of 48kHz, 88.2kHz, 96kHz, and 192kHz, will benefit from our encoding process.
you'll probably get something different, perhaps better, than a CD track ripped to AAC.
Apple is providing the tools they use to convert to AAC so that sound engineers can preview the product before it goes on sale, but they appear to be the same tools they've been using all along. As I said before, "Mastered for iTunes" isn't a new encoding tool - it's a process workflow. Other recommendations:
- Apple recommends listening to your masters on the devices your audience will be using
- Be Aware of Dynamic Range and Clipping
- Master for Sound Check and Other Volume Controlling Technology
- Remaster for iTunes [That is, they suggest starting over from the original recordings, rather than send in a file that was mastered with CDs in mind.]
"Mastered for iTunes" is indeed optimized for iTunes: it's optimized for separating the gullible from their money.
Oh, the beautiful gloss of greality!
Using RHCP records as a basis for comparison is a terrible example; everything they've brought out since One Hot Minute has been overcompressed to death at multiple stages in production (Californication is even cited as a specific example of a crappily mastered record in the Wiki article).
Shortly after reading this article on ars I went to check it out for myself. Yes, technically they are still "just" 256k VBR AAC files just like other stuff in the iTunes Store. But if the engineer doing the mastering has busted his/her ass to play the cat & mouse cycle of re-tweaking the dynamics after listening to the encoded result a few times, the results are extremely surprising.
If you've got a good stereo or a nice pair of headphones, go listen to a normal CD version of Jimmy Smith's "The Cat" ripped at 256k VBR AAC, and then listen to the "mastered for iTunes" version. I had no idea lossily compressed audio from 40+ year old analog master tapes could sound that good.
Correct, FLAC is not inherently "CD quality" — it's more flexible.
CD quality is (two channel) 16 bit samples at a 44.1 kHz rate. FLAC can handle CD quality because it handles a range from 4 to 32 bit samples and frequencies from 1 Hz up to about 650 kHz stored in up to 8 channels. So FLAC has your quality range covered. You could not want for a more capable format. (I think the frequency range means you could encode a substantial chunk of commercial AM radio signal?)
As others note, this the capability range of FLAC, not necessarily the quality of the source signal you're encoding.
FTFAS:
If there are no differences between a Mastered for iTunes format song and a CD version of the same song, what's the bitch? Does he want better quality from a downloadable version?
the Nyquist limit that says you can encode at twice the sample rate applies when encoding a single sine wave.
Or a sum of sine waves, so long as the sample rate is greater than twice the highest frequency. (You need the twice to capture the phase as well as the amplitude.)
I have not heard any music recently that is made entirely of pure sine waves.
People building on the work of Fourier have showed that all music is made of pure sine waves, at least over the short term. Audio codecs divide the signal into "frames", each roughly as long as a video frame, which overlap slightly. Over such a short time period, the fundamental frequency and overtones of each instrument look like pure sine waves. The "underwater" effects in early MP3 codecs came from faulty assumptions about how to deal with transitions from one frame to the next.
An often made error in sample rates involve confusion between sample rate and bitrate. Bitrate is bits per second such as the rate your dial up modem can send including start/stop/parity bits with data bits. This should not be confused with sample rates. Sample rate is how often a sample is captured. For the often touted CD Quality claim of 44.1 Khz, remember that a sample is 16 bits per channel. This is a 32 bit sample taken 44.1K times per second. This does not include any framing, error correction (ECC), encoding (EFM Modulation) etc needed to support Uncompressed audio.
Without any overhead the raw data bit rate for a CD is 44.1K samples per second Times the sample size of 32 bits. This is 1411200 bits per second.
The truth shall set you free!
It is pretty pathetic that iTunes can't even get CD quality. CDs which have been available since the 80's. Why the fuck aren't people selling music with higher quality? There used to be DVD-Audio which was way, way better than CD quality but ever since iTunes all music quality has gone to shit. Give me better fucking quality wtf it is 2012 and we are still using fucking blades on our skin to shave. We should have been using lasers in the 70's and perfected neural implants in the 90's and just will your hair to be shaved or grow with the power of your mind.
The Official Site of 1337 Pwnage
In this video interview with Neil Young, he states that dominant digital audio formats are inadequate and 'some rich guy' needs to build a new iPod that will play up to 30 albums. I don't think he's heard of flac or the Sansa Clip+ (or MOST Android devices).
Most modern music is brick-walled these days anyway. Loudness is winning over dynamic. Half of the reason I find vinyl more pleasing is because it's usually mastered with a lot taken off the limiter.
Actually, chances are the cheaper the gear the better they will sound, as that's what they are mastered for.
First let me start by saying, lossy compression has absolutely no place in mastering. Capture you samples at the highest resolution possible 24-bit/96Hz is passable, 192Hz is better. If you're mastering music on a machine that can't handle this, then simply take when you're given and be off with you.
The author of this article makes claims about performing an analysis of the Apple Mastered for iTunes method. I didn't like his form of analysis. There are major issues with comparing lossy to lossless audio since we have to decide if we want to perform individual listening tests, which often yield nothing of any interest since most "Super Listeners" generally are pretenders to begin with. Frankly, at 24-bit/96Hz or better using high bit rates (as would be used for mastering) chances are, you don't have suitable speakers or headphones to perform the test reliably anyway. Even the best monitor speakers from Genelec can't reproduce the differences at these levels. Some headphones might be able to handle it in some ranges, but not in all. Therefore we're limited to what is possible with pure data analysis.
In data analysis, we can do things like invert the wave of the output and add the results together to produce a distortion wave. As the author claims to have done, but now comes the issue of deriving anything meaningful from this. Often times, better codecs will produce greater distortion because it's supposed to be "meaningful distortion". As such, think of it like dithering an image to improve results from image compression, this is extremely common when down-sampling before compressing. As a result, filters will produce more accurate AC characteristics of the waves being compressed. This comes at a cost of bit rate consumption, but the end result when the image is decompressed is higher.
Some numbers like SNR (signal to noise ratio) are calculated by deriving a meaningful number to calculate how much of the original signal still exist in the compressed signal. Higher distortion such as what this guy is observing can in fact produce better signal retention.
Next is the issue of attempting to produce the best audio output possible based on what the listener perceives as being superior sound. This is a very common issue in audio compression. AAC originally defined 7 possible branches for compression methods which would be selected per audio unit based on a vendor specific 'Psycho-acoustic model". This would attempt to choose the best balance between bit rate allocation and what should sound best to the user. Often these models just tried all 7 compressions and built a quality metric related to how the audio should sound in the end and chose the best output and moved on. Now, though I could be wrong, I believe I know of 21 possible branches for the compression. While each one has its benefits, for a computer to choose one branch over another requires the computer to have a concept of what sounds best. Rarely is it chosen based on pure wave distortion. Often it's chosen based on concepts like "we can improve the quality of this signal by increasing the high frequency distortion since in those ranges, it wouldn't be reproducible by anything other than piezoelectric elements far too small to be heard."
So, in the end, while he may be actually on to something, his reasoning is likely entirely incompatible with reality. Lossless compression chooses how best to lose based on perceived quality as opposed to pure signal quality. A perfect compression would lose nothing and still chop audio to 1/50th the original size. But in this universe, the best lossless compressions still struggle to perform 2:1 reliably and while it can often average or peak at 4:1 can often also end up increasing the size instead. Most of the old Ella Fitzgerald/Loius Armstrong recording are examples of worst case files.
So, while his analysis was very cute, it was possibly inaccurate.
Now, I'll reiterate the first point.
Audio compression DOES NOT BELONG in a mastering environment. Lossless is ok if there's a
Amplitude and frequency are orthogonal. Changing the frequency of sampling does nothing to the amplitude.
Give me Classic Slashdot or give me death!
"Shepherd compared three digital music files, including a Red Hot Chili Peppers song downloaded in the Mastered for iTunes format with a CD version of the same song, and said there were no differences. Apple or someone else needs to step it up here and offer some true 'CD quality downloads.'""
Surely if there were no differences between the CD version and the AAC file, that means the AAC version _is_ CD quality? Quality loss would be a difference, right?
You would think a master engineer would look into the program a little before making a fool of himself. The program is a set of guidelines for mastering, not a different file formats. A lot of engineers may be using similar guidelines when creating their iTunes submissions anyway. There would be no difference in file format. None is claimed. Anywhere. By Anyone.
To clear up some misconceptions here
1: At the nyquist limit you won't get amplitude (or even nessacerally see the signal at all depending on the phase relationship between the signal and your sampling clock)
2: At any frequency below the nyquist limit you can IN theory recover the amplitude but the closer you get to the nyquist frequency the closer your anti-aliasing and reconstruction filters need to be to perfect and this causes problems of it's own (transform a perfect filter from the frequency domain to the time domain and you will find it is non-causal and has an infinitely long impulse response).
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
Though experiment. Take a 1 KHz tone sampled at 10 KHz, phase shift it 180 degrees as described in article, thats 5 samples phase/temporal shift. On the other hand in the 100 hz band you need to shift 50 samples for the same degree phase shift.
and this is part of why the guy is a total non-authority. He is a "mastering engineer" (where did he get his degree, exactly?) who doesn't know the difference between phase and polarity.
There used to be DVD-Audio which was way, way better than CD quality but ever since iTunes all music quality has gone to shit.
DVD Audio is still around.
DVD-Audio authoring software is available for various OS platforms:
For Linux (and BSD and Solaris),
http://dvd-audio.sourceforge.net/
For Mac OS
Burn - open source, free
still running on PPC Macs from OS X 10.3.9, also on Intel Macs, a 64bit-version available, too:
http://burn-osx.sourceforge.net/Pages/English/home.html
Minnetonka Disc Welder - commercial
http://www.minnetonkaaudio.com/index.php?option=com_content&view=article&id=70&Itemid=93&lang=en
For Windows
Minnetonka Disc Welder - commercial
http://www.minnetonkaaudio.com/index.php?option=com_content&view=article&id=70&Itemid=93&lang=en
DVD-Audio Solo - commercial
http://www.cirlinca.com/
You'll also need a DVD recorder capable of DVD-Audio and a respective player.
For PCs, LG Electronics and Pioneer used to have such hardware.