[BearcatSandor]

Hey folks,

What is the Apple lossless compression format called?

Did i read right that you could download music from Itunes in this format?

Bearcat

[drive-thru]

I believe it's known as "Apple Lossless".
I don't think you can currently download purchases from the music store in this format (the downloads would be huge), but you can rip your own CDs to lossless.

[Chito]

I believe the ITMS files are AAC.

[drive-thru]

Yeah, AAC at 128kbps. Audiobooks are mono and lower quality (96kbps I think) in order to keep the file sizes reasonable.

[BearcatSandor]

Hmm...I'm not sure 99 cents a track is worth it to me. Given that if you download the whole disc, you might as well have bought it. Given that a lot of my discs have lots of smaller tracks (Tool etc), downloading a 20 track album sux, when i can buy it in the store for about $10.

Not to mention that as an audiophile, the idea of paying for a ;ossy format makes me shudder

I'm in to a lot of jazz and choral.classical discs too. Wow, the idea of downloading the Murcury Living Presence albums in a compressed format....

[drive-thru]

Yeah, it doesn't makes much sense to buy an album for £7.99 (UK) at 128kbps when I can get it for £8.99 at CD quality. I've used the iTMS a few times, I always get the free track each week, but it doesn't have any great benefit if you want to buy a whole album and care about quality.

[bmedina]

Originally posted by drive-thru:
Yeah, AAC at 128kbps.

And also encumbered with DRM.

The Music Store is mostly useful for buying individual songs, rather than entire albums.

[Randman]

Originally posted by drive-thru:
Yeah, AAC at 128kbps. Audiobooks are mono and lower quality (96kbps I think) in order to keep the file sizes reasonable.

If you buy direct from audible.com, you can set the level of compression.

[Parky]

Originally posted by BearcatSandor:
Hmm...I'm not sure 99 cents a track is worth it to me. Given that if you download the whole disc, you might as well have bought it. Given that a lot of my discs have lots of smaller tracks (Tool etc), downloading a 20 track album sux, when i can buy it in the store for about $10.

Not to mention that as an audiophile, the idea of paying for a ;ossy format makes me shudder

I'm in to a lot of jazz and choral.classical discs too. Wow, the idea of downloading the Murcury Living Presence albums in a compressed format....

But you can buy the whole album for $9.99 you don't pay $0.99 each when you buy the whole album.

CD music is already compressed as all the music is not recorded when it's converted to a digital recording. If you are an audiophile then of course the iTunes tracks will probably not be for you, but to be honest I have been more than happy with the quality for my iPod and streaming to my Airport Express.

Ian

[bmedina]

Originally posted by Parky:
CD music is already compressed as all the music is not recorded when it's converted to a digital recording.

Please elaborate. Maybe you're thinking about dynamic compression? The music on a CD is considered "uncompressed" audio in relation to this discussion.

And while it is true that not all music is recorded on a CD (more precisely, not all frequencies are captured), all the audible sound is.

[BearcatSandor]

Parky,

I think you're talking about the kind of compression that is usually done, post-production.

You can hear it when you listen to most pop cd's and those that are engineered for raidio play.

On a really good system such as the Meridian DSP6000s (*drool*), which retail for $19k a pair, you can hear it. After playing a good Jazz or classical recording you can hear that most pop recordings are missing the lower and higher frequencies and he mid-range is bootsted to make it sound louder and more exiting. It doesn't actually make it louder, your brain just thinks it does unless you retrain your brain.

You can hear an example of this on the SACD release of Dark Side of the Moon by Pink Floyd. On the cd-layer the cymbols drop-off and the bass is muddy. It's almost as if you can hear the space above and below the fequencies that are missing. The midband is boosted to mask this.

Then listen to an original CD or the Moble Fidelity gold reissue and you'll hear a major difference. The bass is tight and the cymbols shimmer.

Listen to the SACD layer and that's another heaven altogether!

Reading this, can you tell why i might not be excited about a download system that features mp3-like audio.


Yes,I am an audio snob.

[Spheric Harlot]

Originally posted by bmedina:
And while it is true that not all music is recorded on a CD (more precisely, not all frequencies are captured), all the audible sound is.

That, too, is not strictly true, since a 44-kHz DAC (digital-analog converter) will introduce measurable aliasing and distortion down to 8 kHz due to the way D/A conversion works.

The primary impetus behind moving to 96 kHz and higher sampling frequencies is to move these unavoidable digital artefacts up above the audible range.

And yes, I think Parky got confused between dynamic compression and lossy DATA-REDUCTION compression.

-s*

[bmedina]

Originally posted by Spheric Harlot:
That, too, is not strictly true, since a 44-kHz DAC (digital-analog converter) will introduce measurable aliasing and distortion down to 8 kHz due to the way D/A conversion works.

Only if you're sampling a signal that hasn't been low-passed at around 20kHz. But that would be unintelligent.

[mattsgotredhair]

I know this is a little late in the discussion, but for me, if I'm paying $.99 for each song, I want .wav's or .sd2's available and the ability to download an entire mp3 preview before i later download the higher quality files. I think that it's crazy that they don't offer anything higher than 128, cause personally, I think 99 cents is way too much for .mp3's. any opinions?

[bmedina]

Originally posted by mattsgotredhair:
I know this is a little late in the discussion, but for me, if I'm paying $.99 for each song, I want .wav's or .sd2's available and the ability to download an entire mp3 preview before i later download the higher quality files. I think that it's crazy that they don't offer anything higher than 128, cause personally, I think 99 cents is way too much for .mp3's. any opinions?

Well, they're not mp3's; they're AAC files from the best AAC encoder (publicly available, anyway). And if the entire song is available before the purchase, people have very little incentive to pay for the song (keep in mind most people don't care about quality above a certain threshold).

People with your qualms are better off buying a physical CD.

[Spheric Harlot]

Originally posted by bmedina:
Only if you're sampling a signal that hasn't been low-passed at around 20kHz. But that would be unintelligent.

I was talking about the D/A stage, not the A/D (sampling) stage.

[bmedina]

Originally posted by Spheric Harlot:
I was talking about the D/A stage, not the A/D (sampling) stage.

I'm only familiar with aliasing as a result of too infrequent sampling. Do you have a link? Surely if that was a problem you would just supersample to a higher hertz before D/A.

[mattsgotredhair]

Ouch, people with my qualms? Damn, I was just saying that if I'm paying as much as I would for an actual cd, I'd like its quality to be as good. Sorry if I some how offended you.

Originally posted by bmedina:
People with your qualms are better off buying a physical CD.

[bmedina]

Originally posted by mattsgotredhair:
Ouch, people with my qualms? Damn, I was just saying that if I'm paying as much as I would for an actual cd, I'd like its quality to be as good. Sorry if I some how offended you.

I didn't take any offense. I'm just saying, if you want CD quality, you should buy the CD.

[Spheric Harlot]

Originally posted by BearcatSandor:
Not to mention that as an audiophile, the idea of paying for a ;ossy format makes me shudder

I'm in to a lot of jazz and choral.classical discs too. Wow, the idea of downloading the Murcury Living Presence albums in a compressed format....

Well, I certainly hope you didn't buy the CDs, then.

[Spheric Harlot]

Originally posted by bmedina:
I'm only familiar with aliasing as a result of too infrequent sampling. Do you have a link? Surely if that was a problem you would just supersample to a higher hertz before D/A.

I'm sorry, yes, you are correct. Aliasing (folding) affects the A/D stage. It's been a while... IIRC correctly, though, to avoid folding by proper filtering, you must sample at roughly 4x the highest frequency to be recorded. So a 96 kHz recording can theoretically be clean to about 24kHz.

However, due to quantization errors, signal distortion can happen far below the Nyquist frequency. A single wave that runs out of phase at some just-barely-off divisor of the sampling frequency WILL go rectangle somewhere and introduce some nasty artificial overtones. This can and does happen down to about 8 kHz.

Here is some detail (excellent resource, btw.) : http://www.sfu.ca/sca/Manuals/ZAAPf/...ion_error.html

[bmedina]

Originally posted by Spheric Harlot:
IIRC correctly, though, to avoid folding by proper filtering, you must sample at roughly 4x the highest frequency to be recorded. So a 96 kHz recording can theoretically be clean to about 24kHz.

You're thinking of Nyquist's theorem, and the factor is 2x. That's why CDs are 44.1kHz, since that safely gets you up to 20kHz, encapsulating the full audible spectrum.

However, due to quantization errors, signal distortion can happen far below the Nyquist frequency.

That's why CD audio is 16-bit, avoiding the quantization noise that occurs at lower bit depths.

A single wave that runs out of phase at some just-barely-off divisor of the sampling frequency WILL go rectangle somewhere and introduce some nasty artificial overtones.

That's only a problem very near the Nyquist frequency and only with very unintelligent D/A stages. And even if the D/A puts out a 20kHz rectangle or triangle wave, your ear will only hear a clean sine wave, since all the higher frequencies contributing to the wave will be inaudible.

In my view, higher resolutions and bit depths are only important during the mixing stages of music production; it's good to have some headroom when you're adding signals together. But for the final mixdown for distribution, 16-bit, 44.1kHz audio is perfect to our ears.

[Spheric Harlot]

Originally posted by bmedina:
You're thinking of Nyquist's theorem, and the factor is 2x. That's why CDs are 44.1kHz, since that safely gets you up to 20kHz, encapsulating the full audible spectrum.

Wrong.

The 22.05 kHz defined by the Nyquist theorem is the THEORETICAL maximum.

In real life, you start getting artifacts above about 8 kHz.

Quantization errors happen both vertically (dynamic resolution, i.e. bit depth), and - much more importantly, horizontally, in the time domain, since a sound wave is a function of time.

You know the effect when you tune a bass to a guitar - one is an octave lower, with a different overtone structure, but you still get that phasing effect, where waves are cancelled or added with a regular period. Imagine the guitar as the sampling frequency.

What happens to the bass tone's waveform as it is sampled and gently shifts around the sampling frequency?

This sort of problem is measurable - and audible - down to about 8 kHz. According to my acoustics prof, the ideal "safe" sampling rate is about 4x the highest frequency to be recorded.

The higher bitrate pretty certainly has the greater effect upon sound quality, though. Even on a simple, straight two-track recording - uncompressed and with some ambience. The difference is there. Of course you are going to master a final production to 16/44.1 until higher-resolution media are common and working with the huge increase in data produced becomes practical.

-s*

[bmedina]

The easiest way to see it is to actually draw out the waveform and see where it's sampled. When you do, you'll see that it's not a problem except near the Nyquist frequency, which is already so high that it doesn't matter.

The phenomenon you're describing is called beating and has to do with the addition and subtraction of waves. Because the sampling "wave" is not added to the captured signal, beating does not occur.

If you have a link that proves me wrong, I'd love to see it. I've never heard of the problem you describe in any signal theory class.

[Spheric Harlot]

Originally posted by bmedina:
The easiest way to see it is to actually draw out the waveform and see where it's sampled. When you do, you'll see that it's not a problem except near the Nyquist frequency, which is already so high that it doesn't matter.

The phenomenon you're describing is called beating and has to do with the addition and subtraction of waves. Because the sampling "wave" is not added to the captured signal, beating does not occur.

If you have a link that proves me wrong, I'd love to see it. I've never heard of the problem you describe in any signal theory class.

The reference to beating was an analogy. Surely it makes sense that as the zero points ("Nulldurchgänge", sorry) of a waveform shift in relation to the sampling frequency (as with two beating tones), there will be a stretch of time where you will be sampling zero points followed by a gradual shift towards sampling maximums and back to zero points. Am I talking out of my ass?

I am quite certain of the 4x sampling frequency rule, but your insistence is casting doubts on the rest, as it has been a few years. I'll be consulting my uni notes.

-s*

[bmedina]

Originally posted by Spheric Harlot:
The reference to beating was an analogy. Surely it makes sense that as the zero points ("Nulldurchgänge", sorry) of a waveform shift in relation to the sampling frequency (as with two beating tones), there will be a stretch of time where you will be sampling zero points followed by a gradual shift towards sampling maximums and back to zero points. Am I talking out of my ass?

No, but as I said, I've never heard of such a problem. And as I said, it's only theoretically a problem near the Nyquist frequency. At all lower multiples of the Nyquist frequency (and, in fact, at all lower frequencies), you get intermediary sample points that are not at the zero or maximum points of the wave.

I am quite certain of the 4x sampling frequency rule.

Here's a link from that website you posted: http://www.sfu.ca/sca/Manuals/ZAAPf/n/nyquist.html