A Look at Apple's 'Mastered for iTunes' Program and its Effect on Sound Quality

[nuckinfutz]

Magnus that's a well versed and tried and true argument that is certainly realistic however there are many examples where a higher bitrate and word length deliver superior sound.

Today's pop music is compressed and then delivered at maximum output that 16-bit playback can offer. At this point I don't really care if it's 24-bit or less.

However if you have say a Jazz or vocal song great dynamics and points where the amplitude is low you're going to like the lowered noise floor of a higher 20 or 24 bit recording.

Plus you avoid the low level issues with 16-bit recording where linearity drops off in the least significant bits. With 24-bit you have more room to avoid clipping, keep the noise floor low and increase the resolution of the LSB.

Can most people hear the difference with average equipment? Likely not as well but I'd rather over-engineer here than under.

cheers.

 

[iBug2]

Well, I took this test once and I guessed 13 of 15 samples right. It depends on quality of mastering, your audio equipment, how experienced listener you are, etc. I believe my results would not be as good as they were if I used some low-end cheapo earphones...

13 out of 15 is quite well still. I'm an experienced listener but not a professional. A mastering/recording engineer will obviously be more experienced than most other people.

 

[iBug2]

Possibly. How could we do the test - I'd almost need someone controlling the test with a set of decent studio monitors and play short 20-30 second segments or something.

Create two versions of the same track. One lossless, one lossy at whatever bitrate you want. Make a playlist consisting of only those two tracks. Select "shuffle" in your player (whatever player you are using, they all have some random function). Put on your headphones, close your eyes and hit play. One of them will be playing. Guess whether it's the lossy or the lossless one, and open your eyes to see if you guessed correctly. It makes sense to listen to both the lossy and the lossless one once before doing this to get a baseline to compare during the tryout.

Do this 40-50 times and check the percentage of your correct hits. If it's above 90% or so, I'd say you can differ between them.

 

[colinwil]

Some poor quality mastering in iTunes Match

This reminds me...

Usually the songs in iTunes match are pretty good. But I'm coming across some songs that sound absolutely dreadful when compared with my original rips. Even old, Mp3 128k ones.

Especially old Rock 'n Roll for some reason. I don't know if anyone else has C'Mon Everybody! by Eddie Cochran? When I ripped it originally it sounded Ok. Not hi-fi of course, because it's old, 50's mono. But still quite vibrant. But now when I play the 'matched' version in the cloud it sounds like someone dipped my headphones in custard.

Have a listen. You'll see what I mean

 

[iBug2]

Well, I took this test once and I guessed 13 of 15 samples right. It depends on quality of mastering, your audio equipment, how experienced listener you are, etc. I believe my results would not be as good as they were if I used some low-end cheapo earphones...

Did you use earphones or monitors? I usually listen with Westone 4's, which are good enough for listening but I'm not sure if they would be enough to differ between lossy and lossless. My studio monitors are Focal Solo BE's.

 

[BlackMangoTree]

The amount of ignorance in this thread is amazing. So many clueless people regarding audio.

Apple will love all the money you send their way.

 

[Owen Imholte]

Possibly. How could we do the test - I'd almost need someone controlling the test with a set of decent studio monitors and play short 20-30 second segments or something.

Create two versions of the same track. One lossless, one lossy at whatever bitrate you want.

Start with that, but then use Juxtapose to compare A and B to a blind "X" a number of times.

There is a video explaining how on my site, but here are the basic steps:

1. Drag and drop the lossless and lossy files onto A and B slots.
2. Click 'start test' and listen to the "3" tracks (switch with the buttons on the bottom or your 1,2,3 keys)
3. Try to identify if X is matching A or B (choose your answer with the buttons on top or hitting A or B keys)
4. View your success rate by clicking "Show Results" (Only do this when you are 'done' or you risk biasing your results)
5. Repeat the previous steps for a while (if you want to see how you are doing, make sure you decide to run a certain number of tests ahead of time, or your results may be inadvertently biased)

This way you won't have to close your eyes and your running score is automatically recorded. Juxtapose is free, the download link is on the right side.

 

[Svirchev]

Range of hearing

"we can't hear over 20kHz so it's just wasted bits"

Sorry, I must disagree with the fundamentals of this assumption. There's these phenomena called overtones and undertones. Experienced listeners can definitevely hear under- and -overtones that less experienced/interested listeners cannot. An audiometric test would rate my top end hearing as significantly <20KHz, yet I can psycho-acoustically feel the overtones way above 20kHz from a musician like Evan Parker or Steve Lacy on soprano saxophone or Francois Houle on clarinet. The earliest CDs (I recall) had a cut-point of 16kHz: they have "missing sound" for any experienced listener of classical, jazz and stuff other than the most mundane pop we are fed.

I recall introducing a friend to some jazz and she could not distinguish a trombone from French horn, but that anecdote tells us nothing about what should be aimed for achieve quality sound. Yes there is diminishing return for mastering for the "average" listener, but if we accepted that line of thinking, then why not just stick with the pre-33 rpm LP dynamic range?

 

[ChrisA]

Vinyls rocks, oh Sound Snob.

No. You should say "Vinyl CAN rock." Some of it is very good and some not as good as the CD version.

It is the same with other formats. 24/96 maybe not be better then 16/44.1 or it might be much better. I have examples of each. I LOT depends on the rest of the recording, the miss used the preamps and room itself and 1000 other things I'm sure.

 

[Winter Charm]

Unfortunately until "Mastering Engineers" stop with the "Loud is good" ********, none of this matters.

Its literally the practice of polishing a turd.

I absolutely agree. If you read the entire article, it explaines how the instructions for using Apple's blend prevent you from doing that, and improve quality, detail, and feel.

 

[ChrisA]

"we can't hear over 20kHz so it's just wasted bits"

Sorry, I must disagree with the fundamentals of this assumption. There's these phenomena called overtones and undertones. Experienced listeners can definitevely hear under- and -overtones that less experienced/interested listeners cannot. An audiometric test would rate my top end hearing as significantly <20KHz, yet I can psycho-acoustically feel the overtones way above 20kHz...

You can explain this with beat frequencies. For example if your speakers can play a 20KHz signal and a 21KHz signal your ears will hear a 1KHz signal. No voodoo at all, anyone who has tuned a musical instrument by ear knows the effect well.

Of course you can't hear the 20 KHz from your tweeter unless you are a dog or under about 16 years old bt you hear the 1KHz soon easy

 

[knucklehead]

Create two versions of the same track. One lossless, one lossy at whatever bitrate you want. Make a playlist consisting of only those two tracks. Select "shuffle" in your player (whatever player you are using, they all have some random function). Put on your headphones, close your eyes and hit play. One of them will be playing. Guess whether it's the lossy or the lossless one, and open your eyes to see if you guessed correctly. It makes sense to listen to both the lossy and the lossless one once before doing this to get a baseline to compare during the tryout.

Do this 40-50 times and check the percentage of your correct hits. If it's above 90% or so, I'd say you can differ between them.

Best to use ABX software to do the test. Unfortunately, good ABX software for OS X is hard to come by. Perhaps there's some better now, but the best I found a couple of years ago was foobar2000 with the ABX plugin. Essential to have the volume level of the samples exactly the same.

Edit-

Sorry Owen. Didn't mean to diss you there.
I haven't tried your software yet. Just glancing at your suggested features list made it look to be in a pretty early stage.
Are you planning on continuing development?

 

[makinao]

Ok but that's because you listened to it thousand times. How about something you listened only once? Do you think you can differ between AAC and 24/196 in a A/B blind test?

But a listener will listen to a piece maybe a hundred times in his/her lifetime, specially one he/she likes. Any shortcoming in the recording and/or performance will become more apparent each time it is played.

 

[ChrisA]

One problem with A/B testing this way. Many times the lossy MP3 files will be quite close except for a few short duration transient effects. The kind of things MP3 has trouble encoding are electronic drum hits and some synth sounds.

The better way to compare is just to listen to the MP3. You may have to listen to a few songs before you hear an "artifact" but when you do hear one you don't need special software it will be just plain obvious.

A/B testing can show any general degradation but those effects are subtle and require good playback gear and an experienced listener.

But "artifacts" are radiaclal changes to the sound and they are common in pop music. Not so common is acoustic jazz or classical.

There is a video explaining how on my site, but here are the basic steps:

1. Drag and drop the lossless and lossy files onto A and B slots.
2. Click 'start test' and listen to the "3" tracks (switch with the buttons on the bottom or your 1,2,3 keys)
3. Try to identify if X is matching A or B (choose your answer with the buttons on top or hitting A or B keys)
4. View your success rate by clicking "Show Results" (Only do this when you are 'done' or you risk biasing your results)
5. Repeat the previous steps for a while (if you want to see how you are doing, make sure you decide to run a certain number of tests ahead of time, or your results may be inadvertently biased)

This way you won't have to close your eyes and your running score is automatically recorded. Juxtapose is free, the download link is on the right side.

 

[AppleScruff1]

Once you go vinyl you'll never go back.

I won't argue that point. I was keeping it digital. Vinyl played on a quality system has a unique appeal and a warm sound.

 

[stationstops]

I thought Mastered for iTunes was designed to take advantage of a legal loophole for record labels to extend copyright ownership of digital distribution into the future.

 

[ramuman]

I'd gladly pay 1.49 or 1.99 per track and 14.99 per album if it were any of the lossless codecs. I have a Mcintosh 402 and Revel Ultima 2s. The impulse to buy a song is always tempered by knowing I'll be able to tell the difference. I'm probably a very small minority for Apple.

It's sort of odd considering Jobs listened to music at home on Wilson Alexandrias.

 

[sonare]

Unfortunately until "Mastering Engineers" stop with the "Loud is good" ********, none of this matters.

Its literally the practice of polishing a turd.

+++ NEWS FLASH ++++

Most ME's hate to have to squash the music -- but guess who wants L-O-U-D-E-R? The clients and record companies, and they are cutting the check. I am lucky in that my clients trust my judgement. Only ONCE have I been asked to make it louder, and I simply responded, "have you compared your disc to the same genre CDs from major labels?" The whining stopped--

Rich

 

[petsounds]

Hello all,

I'm sorry, but you guys who are meow meowing about not having ALAC stuff... while I understand why you guys want it, 99% of customers don't want the huge files, and Apple doesn't want your iDevice getting too full. Both inhibit sales.

Oh, and I bet if anyone took a Pepsi Challenge on 256 AAC vs 96/24, you would be statistically random. I can beat a 320 mp3 every day of the week, but I can't beat 256 AAC. If that's the case, why can't 256 AAC satisfy? It's not like we all run around saying OMG I NEED UNCOMPRESSED VIDEO - usually people are pretty satisfied with BluRay - hell, even 4k is compressed, but you WANT ALL THE BITS in audio? Come on. Are you the same guy that says you want it on vinyl because that's better? Gag.

Your first paragraph is a non-issue. Apple could transparently transfer compressed AAC files when songs are copied to iDevices. If PS3 Media Center can transcode 1080p video on-the-fly, iTunes should be able to encode audio on-the-fly no problem.

I've taken the Pepsi challenge, and can pick the 24-bit files versus the 16-bit files. 96 vs. 44.1 is more debatable, but the higher bits-per-sample is absolutely detectable if you were blessed with great ears. I've talked to many audio engineers who agree on this point. Please note that this is with a high-quality Digital-to-Analog converter chip. The iDevices are alright, but don't hold up to those found in higher-end home stereo equipment (or pro audio gear). With crappy DACs you can get sample dithering and jitter that makes everything sound brittle and hollow, no matter what quality it is. I haven't done any blind tests with an iPhone; I think their current chip only does 24-bit / 48kHz.

I would also note that iDevices aren't just being used for jogging/gym/car now. People are using AirPlay to stream to their home theatre, and thus the the quality of files becomes more pertinent. This is complicated by the fact that the Airport Express's aging DAC doesn't support 24/96 streams.

 

[JAT]

Challenge accepted

Any non-damaged HDMI or TOSLINK cable will transfer a perfect signal from source to processor. Any decent set of RCA cables will transfer that signal to an outboard amp, if involved. Any acceptably thick strand of copper wire will pump the signal from the amps to the speakers.

I'll just bring up 2 things if you'd like to play with the wire crazy.

1) HDMI cables are NOT all the same. First, they are not all the same gauge. That right there is enough to completely change the discussion. And I didn't even mention length!! Also, digital signals are not nearly as easy to transfer as audio. So all the ranting that people do about "acceptably thick strand of copper wire" does not apply equally to audio, video, and digital. Yet, in internet audio arguments that seems to be forgotten. Repeatedly.

2) You used the word "acceptably". You sound like an audiophile.

 

[JAT]

*sigh*

I was fine with everything you said, and then you said this.

Take a look at these two numbers: 312 312

Can you make out the difference? That's the literal difference between an error-free ALAC and the original AIFF. Let me know when you can make out the difference.

Are those numbers for engineering? Accounting? Math? Digital audio?

Let me put it another way....Do you want them to be equal? Or do you want them to have differences?

 

[SDAVE]

Me too...sell me a NON-APPLE PROPRIETARY LOSSLESS FORMAT (such as WAV, APE, FLAC) and I would buy hundreds of songs a year. I own 0 from iTunes. I have over 24,000 songs...all ripped from my cds that I purchase each week from Amazon below $10 a pop with free shipping and 0 tax.

Apple Lossless is open-source now and many free tools out there support it without issues.

 

[derbothaus]

*sigh*

I was fine with everything you said, and then you said this.

Take a look at these two numbers: 312 312

Can you make out the difference? That's the literal difference between an error-free ALAC and the original AIFF. Let me know when you can make out the difference.

One is to the right and one is to the left.
Bit for bit yes but it is still a compression algorithm. So it is technically different. Maybe mp4 container holds unpleasantness? Kidding. Playback can err and iTunes versions can sound different even though it is playing back the same files.
There was a reason I said I could not tell the difference. Love the *sigh*, you're awesome.

 

[Johbremat]

I regret ripping all my CD's years ago and tossing them. Had I known better I might have better quality tracks.

Don't they call that piracy, and karma?

 

[Lynn Belvedere]

Hello all,

I'm sorry, but you guys who are meow meowing about not having ALAC stuff... while I understand why you guys want it, 99% of customers don't want the huge files, and Apple doesn't want your iDevice getting too full. Both inhibit sales.

Oh, and I bet if anyone took a Pepsi Challenge on 256 AAC vs 96/24, you would be statistically random. I can beat a 320 mp3 every day of the week, but I can't beat 256 AAC. If that's the case, why can't 256 AAC satisfy? It's not like we all run around saying OMG I NEED UNCOMPRESSED VIDEO - usually people are pretty satisfied with BluRay - hell, even 4k is compressed, but you WANT ALL THE BITS in audio? Come on. Are you the same guy that says you want it on vinyl because that's better? Gag.

You sound completely uneducated on the subject. 256 aac = VHS, not Blu-Ray.

Selling lossy audio in 2012 is a complete joke. 16/44 ALAC (CD standard) should be the absolute minimum for inclusion in the itunes store. If you decide to downgrade that audio to fit 5000 songs (most of which you never listen to) on your ipod at once to make you feel cool, then that is your choice. But to actually sell this downgraded, ultra-compressed, tinny-sounding garbage at full CD price is a farce.

People who buy music on itunes are complete suckers.