iTMS: Satisfied with bit-rate?

I don't know for sure what the encoding is like on the previews. There's absolutely no reason for us to think that it's at 128kb/s because there's been no indication anyway. I've listened to some previews and I'd imagine that they are less than that, just because they don't sound great. I've bought a couple albums and I can't complain about the quality. I bought the coldplay album and previously I had only heard it at 160 mp3. I think the aac version is better.

edit: let me also add that the other album I bought was Coltrane's Blue Train and I haven't noticed any problems in the encoding (using my decidedly non-audiophile ears and a good set of headphones).

I've been pretty thrilled. The few tracks I purchased so far I had previously only heard on cassette so the 128KB files are stunning in comparison .

I just listened to this same passage. I don't have the same music in any other format for comparison, so I can only say that in isolation it sounded just fine to me. The only thing that sounded like warble to me was a little vibrato. Certainly no part of it sounded harsh or tinny.

Can you somehow get a hold of the uncompressed source from a CD for comparison? Maybe your MP3 is lacking some audible content that the AAC version doesn't miss, but you've gotten used to the sound of the MP3, so that version sounds better to you.

No I heard just that opposite. The Previews may be at 128kbps but they are smaller for quick previews. AAC has different levels of compression "Profiles" of some sort.

I've heard numerous reports on maccentral.com about anomalies in the Previews that don't show up with the downloaded file.

I'll rip clapton from CD to a few different formats and play it through my stereo. I wonder if the file system on my CD will read MP3 ad AIFF off the same disc. Just so I can't tell, hear discs switching etc etc...

I could run AAC from the mac to the reciever, but then they wouldn't be using the same source.

My impression in the past has been that low bit rates like 128-192 certainly are noticeable, but not unpleasant at 192. At 320, I don't notice too much difference unless I play it back to back against the CD.

Have you done any extensive blind testing of this? I thought I had pretty good ears for this kind of thing, but did terribly in a 192 vs 320 test not too long ago. Damn placebo effect.

Um, your Mac can play CDs and AIFFs you know

The mac most surely doesn't do as good a job as a dedicated CD player, and it probably doesn't do as well as my DVD/MP3 player either. Computers usually make a shitty source that will mask many inadequacies of a recording. Well, that's not true, they're up to Mid-fi quality.

Just trying o get a like with like comparison. Same system, same source, differnet formats. Though this isn't anywhere near as rigorous as it needs to be. My DVD player doesn't play back AAC either.

My buddies nephews make CD's for his car using MP3's converted to AIFF. They sound crappy, very muddled.

If the point of this tech is that we can make CD's for our use, encoding them using a 128kbps source is a recipe for dissapointment.

Are you saying that the AIFF sounds crappy, just like the MP3, or are you trying to say that the AIFF sounds worse than the MP3 from which it was derived?

If the latter... that makes no sense. AIFF is just a storage format for uncompressed (or in this case, decompressed) digital audio. Unless you're changing sampling rates along the way, or something unusual like that, an AIFF file contains the same information which gets streamed into a DAC, and then to your ear, as when the original MP3 is played.

There is generational loss when encoding or reencoding into a lossy format. There is no generational loss from intermediate uncompressed encodings (barring anything unusual like changing sample size or sampling rate along the way).

I'm saying that his nephews steal crappy 128-192 Mp3's on Kazaa, and they encode/export them to AIFF, to make CD's that will play in any CD player, and that those CD's sound muddy/muddled, like crap basically.

If part of the Appeal of this service is that I can encode the tracks back to AIFF for playback in a (non MP3 capable) stereo, then we need at least 320Kpbs to make a passable transition from MP3 back to AIFF.

Make sense?

This distortion is not necessarily caused by the compression in AAC. More likely, it's a function of the iTunes "Sound Enhancer" and/or "Sound Check" features. When these features are enabled, iTunes reads the digital recording, it looks for spikes in frequency and volume, and attempts to smooth them out.

Most of the time this is desirable. Because recordings can vary so widely in volume, etc, you could be listening to your library on shuffle and constantly have to be adjusting the volume. However, if you've got a pair of higher-quality speakers or headphones than most people playing music off a computer, you're more likely to notice these artificially imposed cielings.

Try turning off one or both of those options in iTunes Preferences, and see if the music sounds better to you.

If the mp3 sounds crappy, the AIFF will sound crappy too.

I think that 128bit AACs sound great, and I've compared them to the original CD on my stereo.

But when you burn them onto CDs, make sure that you are burning them at low speed - 1x preferably.

Every time you encode/re-encode, there will be generational loses in quality. Making an AIFF CD from an MP3 source will not sound as good as making a straight copy of the CD track. Compression doesn't squeeze anything, it throws things out. This might fool our ears, but it doesn't fool your CD player. The resultant CD is in essence mastered from a lower quality source, it's not only as bad as the MP3 it cam from, it's worse! It might sound OK, but the "fidelity" to the original is greatly comprimised each time a recording is transfered to another format using lossy schemes.

It's not like copying an MP3 (bit for bit), those will sound identical no matter how often they're copied. If you re-create an AIFF from an MP3, what was lost in the original encoding is gone. It's re-invented in a sense (by your software) to remake a playable track in whatever format, but the result is even further from the original recording.

True.

Making an AIFF CD from an MP3 source will not sound as good as making a straight copy of the CD track.

Well, mostly true except that a standard CD is not an "AIFF CD" -- it's a Redbook standard CD, and it doesn't store AIFF files. If an AIFF file is encoded at 44.1 KHz stereo with 16-bit samples, the numeric data in the file will, however, be equivalent to what gets stored on a standard CD.

Compression doesn't squeeze anything, it throws things out.

True enough.

This might fool our ears, but it doesn't fool your CD player.

Now what the heck is this supposed to mean?

Let's take it for granted that a given MP3 file has succeeded at "fooling your ears". You can't directly play the MP3 file in its compressed form... it always has to be decompressed before being fed to a digital-to-analog converter (DAC). When you decompress the MP3, it is turned into the very same audio data whether you immediately feed that signal into a DAC, and then into speakers and your ears, or if you write that data to a CD as an intermediate step.

All else being equal at playback time... DAC quality as good, speakers as good... if the MP3 fools your ears, the CD made from the MP3 will also fool your ears.

The resultant CD is in essence mastered from a lower quality source, it's not only as bad as the MP3 it cam from, it's worse!

Worse than the original CD, yes. Not worse than the MP3. If the MP3 succeeded in fooling your ears, the CD made from that MP3 has just as much capability to do so. The CD simply has become an intermediate buffer for the very same data you'd have been listening to when listening to the original MP3.

It might sound OK, but the "fidelity" to the original is greatly comprimised each time a recording is transfered to another format using lossy schemes.

Aye, there's the rub. That second-generation CD might manage to sound as good as a first generation CD, but it makes a poor source for a second round of ripping and encoding. Encoding noise that was below the threshold of human perception the first time around will get added into a new cycle of encoding noise. The total encoding noise might then reach audibility.

I'm listening to some Suzanne Vega from the iTunes Music Store, and it's damn good encoding. i'm of course not listening to this stuff on great speakers, but it's coming across to my damaged ears as being pretty crisp and clear. I did hear some Van Hlaen stuff at the Store that sounded por though, but for the most part, the stuff sounds quite good.

Is it satisfactory? Yes, but I'd be much happier with a bit more leeway. On some tracks you really notice how vocals and instruments munge up each other.

any more leeway and you get into possible problems with piracy... i say if after 3 months of the windows version being out and there is no major piracy problems from exclusive tracks then apple will be allowed to offer better quality rips (I hope they have already planned for this \)

.

Well what do you know, our friends at MSNBC think the sound quality is not quite "up to par." In fact they think that "Apple downloads ring sour note."

Fitting. Considering the article is written for MSNBC.

I don't know how everyone else feels, but I've recently noticed that I can identify AAC encoded files in other people's iTunes shares... This is before I select view options to confirm the type/rate.... And, more alarmingly, this is with music I don't know well.

I can't really express what I'm hearing, it's nothing obvious like the ringing or someone in sodden rubbers trodding all across your music that happens with say.... 128kbps mp3. It's not particularly abrasive, but there just appears to be *something* in the music that makes me say.... that's an AAC... and boom, it IS.

I realize that all that sounds about as convincing as seeing the Virgin Mary in a mound of cigarette ash, but hey, it's what I'm hearing.

I'm sure there's a little alternative motivation behind the articles, BUT, they're not entirely off the mark. For a dollar a song, basically the cost of a pressed CD (when you add up the tracks.) iTMS really ought to offer better quality than 128Kbps. You're getting convenience and flexibility, but not "quality".

Add me to the list of people who can tell the difference between an MP3 and a CD track, rather easily. Tested out a few differnet bit rates of Cohen, Springsteen, Susan Vega, The Doors, Mazzy Star, SRV, just a little this and that, what was in the CD changer, and a quick disc of MP3's from 128kpbs, a bunch of 224-256 and a few 320. You can tell, but the later are not offensive and quite listenable.

Didn't try out AAC yet.

Now I could have some (or all) of this wrong, but converting an MP3 to a CD track should further reduce the fidelity because of way MP3's work. They throw out info, this much we agree on, but what do they do when they play back? This sound must be reconstructed out of somewhere. A CD, while not a compression format, has a finite resolution described by a sampling rate and a word length. Mebbe it's not perfect, but working from a good recording you can do some nice things. Working from an MP3 it has to sample 44.1K 16bit words (1422kbps of data) from a maximum of 320kbps of data. It's like having only one of every fifth bit and having to guess the rest in the conversion process. I probably still haven't explained it well, but this is what I meant by "Not fooling your CD player" These "new" bits are a fabrication, they're plotted against a general guide to what the sound should be, but basically they're another level of guessing. In the MP3 source, sound is merely "missing" in the CD made from said MP3, the "missing" sound has been effectively "estimated/invented" to get the requisite "bits" into the track. It's even at this stage further removed from the original than the MP3, not because you throw out more info, but because you have to invent some to get the fomat to fit. If you then make an MP3 rip of this CD (created from an MP3 source) you create yet another generational loss.

That's the way I understand it, the quality is worse than the Mp3, (probably not noticeably) because now the bits are not only a smaller sample of the original, they're mixed iin with 4/5ths worth of psuedo-bits!

Naturally, if we stay within the realm of MP3, like any digital medium, it can be copied infinitely without a loss in quality, but if we start converting from one format to the other, regardless of the direction conversion, "fidelity" (the sameness to the original recording) diminishes.

This is why, I think that if the goal is to allow me to have a track that I can use to make a "CD" copy for myself, for play in my car, on my stereo, etc etc... then it needs to be the highest bit rate possible, if they're just for computer/Mp3-portable player use, then a slightly lower quality is OK, but by no means ideal.

Now, if you buy good albums from good artists, then the cost of buying all the tracks is still worth it. iTunes is strickly for "one hit wonders" or one or two "singles" from otherwise weak artists.

Quick techie question:

Does burning music onto CDs at lower speeds enhances playback quality on a cd player or just makes CDs more compatible?

Sorry to say, but your first conclusion here shows some confusion about how MP3s work.

They throw out info, this much we agree on, but what do they do when they play back?

MP3's do exactly the same thing, in terms of audio data, when you play them back directly, or when you write them to a CD.

Here's the typical direct playback chain:

MP3 file -> decoding -> 16-bit * 44.1K/sec audio stream -> digital-to-analog converter -> amplification -> speakers/headphones -> your ears

Here's the MP3 burn process:

MP3 file -> decoding -> 16-bit * 44.1K/sec audio stream -> store to CD

When you playback the CD made from the MP3:

read from CD -> 16-bit * 44.1K/sec audio stream -> digital-to-analog converter -> amplification -> speakers/headphones -> your ears

This sound must be reconstructed out of somewhere. A CD, while not a compression format, has a finite resolution described by a sampling rate and a word length. Mebbe it's not perfect, but working from a good recording you can do some nice things. Working from an MP3 it has to sample 44.1K 16bit words (1422kbps of data) from a maximum of 320kbps of data. It's like having only one of every fifth bit and having to guess the rest in the conversion process. I probably still haven't explained it well, but this is what I meant by "Not fooling your CD player"

A digital-to-audio convertor (DAC) has exactly the same demands that you state that a CD has. Your Mac (or your PC or your iPod) does not contain any built-only-for-compressed-audio DAC that can be fed directly from an MP3 file. That very same data that you write when you burn a CD is the same data that gets fed to a DAC so you can listen to an MP3.

By the way, all of this applies to AAC or any other compressed audio format as well as to MP3.

I can hear that a CD burned at low speed sounds better.

Thanks shetline, mebbe I'm a "little" dense here -- or a lot! -- but I'm still not getting it. OK, so I know the DAC converts the signal to something listenable (audio, hence D A C), and then this is recorded to CD. See, I'm haven't explained myself well, but it seems to me that it is impossible for the recorded CD to have more resolution than the source (in this case an MP3) yet the recording of a CD requires 1422bits of info. Sure the DAC plays back an MP3 at 44Khz 16 bit quality, but this is from a real maximum of only 320 bits. The rest of the resolution recorded onto the CD (the other 1102 bits) where do they come from? They have to be recorded from a model that is essentially missing them, or allowing an algorthym to reconstruct them (is this not the psycho-acoustic model of MP3). I think in a sense, it is valid to say that "bits" have been invented or guestimated when you do this.

Ahh, I don't have the lingo, and more than likely a cloudy understanding, someone help?

We're going along fine here, and then...

, and then this is recorded to CD.

The output from a DAC is not what gets recorded to a CD. CDs are digital, and the output from a DAC is analog.

It's the output from an MP3 decoder, which is a decompression algorithm that gets fed into a DAC so you can listen to an MP3, or that gets written onto a CD when you burn the MP3.

You're concerned about missing bits? (That's not exactly the right way to look at it, but it'll do for now.) Even with CDs totally out of the picture, you can't listen to an MP3 until those "missing bits" have been filled in. A DAC needs to have every bit spelled out too, just like a CD, or it isn't going to be able to create a listenable signal.

See, I'm haven't explained myself well, but it seems to me that it is impossible for the recorded CD to have more resolution than the source (in this case an MP3)

No, not more resolution. Equal resolution. I never claimed that a CD made from an MP3 file sounds better than the source MP3. It is equal in resolution to the MP3 from which it is created, and, like the MP3, lower in resolution than the original CD from which the MP3 was derived.

yet the recording of a CD requires 1422bits of info. Sure the DAC plays back an MP3 at 44Khz 16 bit quality, but this is from a real maximum of only 320 bits. The rest of the resolution recorded onto the CD (the other 1102 bits) where do they come from?

The CD and the DAC both get these "missing bits" from the very same place -- the same MP3 decoder that unpacks the MP3 data to feed either a DAC or a CD burner.

They have to be recorded from a model that is essentially missing them, or allowing an algorthym to reconstruct them (is this not the psycho-acoustic model of MP3). I think in a sense, it is valid to say that "bits" have been invented or guestimated when you do this.

I'll try to find a good link later for how MP3 and most other audio compression schemes work. Fascinating stuff! (Well, at least to an appropriately geeky mind. )

I've also heard that painting the edges of your CDs with a green magic marker will make them sound better. That putting a brick on top of your amplifier will make CDs sound better.

Subjective opinions of sound quality are notoriously susceptible to the power of suggestion.

Burning speed effecting sound quality, short of blatantly obvious skipping, popping, and clicking, etc., makes no sense at all if you know anything about the way CDs work.

Of course, many people don't give a hoot whether something makes sense or not. They tried slower burning. They heard an improvement. That settles it. But really, when usually this kind of stuff is just a placebo affect, just how many placebo effects do we need to collect and disseminate in the guise of sage advice over and over again?

Thanks for the tip. I was having lots of problems getting my burned discs (24x) to play in my 4 year old SONY CD changer (car). Sometimes it just wouldn't recognize the disc and maybe skip tracks at best. After reading your comments last night, I tried burning at (1x) with the same burner and same CD-R brand. This (1X) burnt disc works just fine in my car. So, I guess burning speed does indeed have an effect on playback quality and compatibility with most players out there.

I couldn't hear any sound difference when played on my (same car) in-dash CD receiver though.

Thanks,
Bill

This kind of failure from different burn speeds makes sense. Nearly any failure in a digital medium is not subtle.

What needs to be taken with a massively large grain of salt is when someone claims that a slower burn makes a CD "sound better", aside from blatantly obvious clicking, skipping, and outright failure to play.

"The vocals sounded richer and more detailed. There was better localization of the sound stage..." Blah. Blah Blah. Odds are that no matter how certain someone is that they've heard these effects and that they aren't fooling themselves -- they are being fooled. It's just placebo effect, the power of suggestion.

I'm not talking about a richer sound, better localization and so on.

You obviously agree that burning discs at high speed means more errors. Many errors on a CD = bad sound.

If by "bad sound" you mean obviously blatant clicking, skipping, and failure to play, yes. That's definitely bad sound!

But "bad sound" as in some sort of general gestalt quality of the sound -- hissy, tinny, muffled, grainy, poor imaging, etc. -- no, I won't agree there. High error rates will not produce these kinds of effects.

There are three classes of errors when reading data from a CD:

• Recoverable errors: The audio data is not read perfectly, but any errors are detected and corrected, restoring the signal exactly to its original intended form. This can have no effect on sound quality, because you're still recovering exactly all of the audio data in its original form.
  
  Simple experiments have shown that even with cheap CD players the most typical result of playing a CD that's in reasonably good condition is that errors are infrequent, and when they do occur, they are nearly all of the recoverable variety.
  
• Concealed errors: The audio data is not read perfectly, errors are detected, but the proper values for the corrupted data cannot be determined. In this case, the CD player must use guesswork to determine the missing data. Does this decrease fidelity? Sure... but only for a very brief moment. For an over-all unsatisfying sound quality to result from concealed errors you'd have to be getting concealed errors at such an extraordinarily high and continuous rate that you couldn't help but also having plenty of unrecoverable, unconcealed errors at the same time.
  
• Unrecoverable, unconcealed errors: The audio data is so corrupted that errors can no longer be either corrected or concealed. The audible results are blatant: pops, clicks, chirps, skipping, complete failure to play.

It's in the nature of digital media: You get perfect recovery of data, or you get obvious failures. There is the remote theoretical chance of something in between, but the odds against getting such results, especially on a consistent basis ("All I know is that my 1x CDs sound better to me than my 12x CDs!) are astronomically high.

I've read about the CD burning speed issue, and I think it might be dirctly related to the depth of the pits burned on the disc. According to Yamaha, it is possible to burn deeper pits, more like from a pressed CD, and thus more easily read by even older CD players. I haven't encountered a CD player that won't play burned discs, even my slightly older carousel. I imagine that as speeds of media increase, they make a slighter impression on discs. No problem for CD-rom drives, but sometimes the pits and valleys may not be distinct enough for some players, so burning at a slower rate might ensure that each pit is deeper. I forget what Yamaha calls this feature of their burners, but they claim it results in a disc with pits almost as deep as stuio pressed CD. Interestingly, this feature only works up to 4X speeds. iDunno. I would imagine that all bit for bit copies of CD's have IDENTICAL audio quality to the original, and any reductions in quality would have to be directly due to an inability to read the burned media.

Still can't quite wrap my head around the MP3 to CD thing, though. Ah well.

Also CDRs and CDRWs don't burn physical holes:

Just a quick question- is there a difference betweenthe AIFF and CDDA formats? I've tried ripping CD's to AIFF in iTunes and get slightly larger files, but have been told that they are synonymous. What's the real deal?

Ah, an earlier article I read referred to the technology making "pits" and "lands" and even earlier Yamaha graphics suggested "pits" but what they're really playing with is the length of the reflective/opaque dots and dashes. Still within tolerances, but just a bit longer to make a clearer distinction between peak and valley. The result should be playable in more players then, and last a bit longer before it's susceptible to errors.

I've listened to AAC files from both iTunes and AACelerator, at varying bit rates, and compared them to MP3 files in the same bit-rate range. My very general conclusion is that AAC encoded files are about the same size at the same bit rate as MP3, however you can save space because you don't have to go up around 256 kbps or higher to get close to CD quality with AAC.

I don't suspect 128 kbps is really "CD quality"; that's RDF all the way. I do think however, that 128 kbps AAC is easily equivalent to about 224 kbps MP3, maybe 256 on certain types of audio. But it's not as pure as what you get from a CD at this point IMO.

One person mentioned this but no one else commented on it: you should definitely turn off the Sound Enhancer and Sound Check in your iTunes preferences.

I was wondering what the heck was going on when I was listening to my music and in some parts, the music would get quieter... I thought there was something wrong with this AAC stuff, but it also happened on my old mp3s too. So I discovered the Sound Enhancer and Sound Check options and turned them both off. Now the music plays as it should--no funky changes in volume levels...