[USNA91]

I just ripped about 100 CD's, and now I find out that there are numerous options to ripping CD's with iTunes. I'd like to think I chose the best one, but....

Can someone provide, or refer me to a place that provides, a decent explanation about the different formats, bit rates, etc., available through iTunes?

Thanks.

[bmedina]

I'm not aware of a site that lists specifics from an iTunes perspective, so I'll give it a shot.

As far as codecs in iTunes go, you can break them down into two camps: lossless and lossy. Lossless means that the decoded output of a file will be bit for bit identical to the input. Lossy means that the decoded output is different than the input (i.e. you've "lost" or discarded information in the process).

The lossless group includes AIFF, WAV, and Apple Lossless. AIFF and WAV are uncompressed, meaning that they are the same bitrate as the files stored on the CD, namely 1400 kilobits per second (kbps). Apple Lossless is a proprietary format that Apple introduced this past year. It achieves about a 2:1 compression ratio (i.e. the files are about 700 kbps) while still being lossless.

The lossy group includes MP3 and AAC. Lossy codecs can achieve much higher compression ratios (around 10:1) while still sounding very good. They achieve this by discarding information from the waveform that is inaudible (hopefully), making the waveform simpler and thus, easier to encode. Generally, the higher the bitrate you choose, the less information the codec has to discard, and the better the file will sound.

MP3 is the older codec (part of the MPEG1 standard), and Apple's encoder in particular is not known for its quality (quite the contrary, actually). AAC is relatively newer (part of MPEG2 and MPEG4), and Apple has put a good amount of effort into their encoder.

There are many more details to all of that, so let me know if you want more information or clarification about something.

With that background, the next question is what to use: lossless or lossy? This question can spark a huge debate on its own. The issue is somewhat complicated, but if you answer these next questions, I think you can get an answer:

• Are you paranoid about the possibility of quality loss in going from the CD to computer versions of your music?
• Do you need the ability to re-encode your music to different formats in the future?
• Do you hate the idea of re-ripping your CDs and want to have one, perfect copy of all your music on your computer?

If you answered yes to any of those, you may want to use lossless. Again, I can explain further the reasoning behind these questions if you want, but I don't want this post to be (too many) pages long. If you choose lossless, I would start with Apple Lossless. Keep in mind that, by definition, lossless music can be re-encoded to other lossless codecs with no loss in quality, so the choice between lossless codecs isn't essential to begin with.

If you don't care about any of that, then use lossy. The big advantage is size: you achieve much smaller file sizes with lossy codecs. You will prevent your hard drive from filling up quickly, and the files are much more amenable to portable uses (like iPod use).

The question is then, AAC or MP3? Keeping in mind that iTunes' MP3 encoder is pretty awful, the only reason to use MP3 is for compatibility with other devices. If you don't have that restriction, use AAC.

The final step is, what bitrate? iTunes defaults to 128kbps AAC. This is a good trade off between file size and quality. But always keep this in mind: only you can decide what bitrate is good for you. No matter what other people say, you have to experiment for yourself; everyone's hearing is different. My advice is to start with 128kbps AAC. If you find that unsatisfactory, move up to 160kbps, then 192kbps. If you need to go past that, you're into the area of paranoia, which means you should consider going lossless instead.

Finally, a good resource for all things related to audio compression is hydrogenaudio.org. Many developers of audio codecs frequent those forums, so it's an excellent place to lurk and learn.

Sorry for the long post. Please ask questions if something is unclear or you want more information.

[jmiddel]

Thank you, bmedina, for a very thoughtful and informative answer, it clarifies quite a few questions that were on my mind.

[analogika]

Originally posted by bmedina:
They achieve this by discarding information from the waveform that is inaudible (hopefully)

Excellent post, bmedina, but just to clarify:

During lossy compression, you lose quality. By definition. Period.

Whether or not you hear a difference depends entirely upon what equipment you're listening on, and how astute your hearing is. Different types of music/instrumentation may also take to lossy compression better or worse.

-s*

[Busemann]

The widely regarded sweet-spot for AAC is 192kbps with the iTunes encoder, while the sweet-spot for mp3 is "--preset standard" with the LAME 3.96.1 encoder.

In any case, you should stay away from the iTunes mp3 encoder if you are concerned about quality.

[bmedina]

Originally posted by jmiddel:
Thank you, bmedina, for a very thoughtful and informative answer, it clarifies quite a few questions that were on my mind.

I'm glad you found it useful.

During lossy compression, you lose quality. By definition. Period.,

But quality itself is subjective. I'd say, "During lossy compression, you lose information. By definition." Whether or not that loss of information degrades the quality is another matter.

One more note. Quicktime 7 will include AAC VBR (variable bit-rate) encoding. If you're starting now to rip to AAC, it may be worth waiting to see how well the VBR mode compares to the current constant bit-rate encoder.

[analogika]

Originally posted by bmedina:
I'm glad you found it useful.

But quality itself is subjective. I'd say, "During lossy compression, you lose information. By definition." Whether or not that loss of information degrades the quality is another matter.

No.

Whether *you*, personally, can discern the loss of quality is the question.

Less information == lower quality.
Nothing subjective about that fact.

That's just nitpicking, though - if he and you find a lossy codec setting where you are unable to tell the difference or simply don't care anymore, then the quality is, for all intents and purposes (EXCEPT further transcoding at a later stage!), "good enough".

-s*

[bmedina]

Originally posted by analogika:
Less information == lower quality.
Nothing subjective about that fact.

Ok, so let's say I have a waveform sampled at 96kHz containing a signal at 30kHz. I feed it to a lossy encoder, and it low-passes at 22kHz, cutting out the 30kHz signal. It certainly lost information, but did it degrade quality? The reasonable answer is "no", since anything at 30kHz is inaudible. Hence, less information != lower quality.

In fact, a great amount of R&D has gone into the very subject of losing information without lowering quality. It's the whole basis of lossy encoding.

find a lossy codec setting where you are unable to tell the difference

IOW, a setting where the output is subjectively of the same quality as the input?

[analogika]

Originally posted by bmedina:
The reasonable answer is "no", since anything at 30kHz is inaudible. Hence, less information != lower quality.

In fact, a great amount of R&D has gone into the very subject of losing information without lowering quality. It's the whole basis of lossy encoding.

By extension of your argument, there should be no audible difference whatsoever between different lossy encoder formats, since they're all based upon a great amount of R&D to ensure quality.

That is obviously not the case.

No: "Quality" in terms of audio is fidelity to original signal. Anything that is not 1:1 is a degradation of quality. Period.

Once you get into the subject of lossy processes, such as lossy encoding, or A/D conversion, or D/A conversion, or even simply reading audio data from a red book CD (all of which ALWAYS incur a certain loss of fidelity, and thus quality), things may become a matter of taste at some point.

Apart from that, there is considerable evidence that the supposedly "inaudible" high-frequency spectrum above 20kHz actually *does* make a difference. Nobody has been able to pin down the precise psychoacoustic effects yet, but research has shown a noticeable difference in cerebral activity when double-blind test subjects were played recordings with and without the >20-kHz frequencies filtered out. (Reference: James Boyk, CalTech Music Lab)

Just because YOU don't hear a difference in quality, doesn't mean there isn't one.

Hey, people drink Earl Grey teabag tea all the time and think it's great quality...

-s*

[Busemann]

Originally posted by analogika:
By extension of your argument, there should be no audible difference whatsoever between different lossy encoder formats, since they're all based upon a great amount of R&D to ensure quality.

Had all codecs been the result of the same amount of research, the the answer would be "yes". This is obviously not the case, however.

The point with psycho acoustic models is to fool the brain into believing what we hear is "CD quality". Objective listening tests have shown that this can work in real life, as long as you use the right encoders with the right parameters. Some samples are indistinguishable to the original, hence no quality loss.

[analogika]

Originally posted by Busemann:
Had all codecs been the result of the same amount of research, the the answer would be "yes". This is obviously not the case, however.

The point with psycho acoustic models is to fool the brain into believing what we hear is "CD quality". Objective listening tests have shown that this can work in real life, as long as you use the right encoders with the right parameters. Some samples are indistinguishable to the original, hence no quality loss.

No.

PERCEIVED QUALITY != QUALITY.

Lossy formats LOSE AUDIO QUALITY. Period.

Whether you can hear it or not.

Loss in resolution == Loss in quality. That is so, REGARDLESS OF WHETHER ANYONE, ANYWHERE CAN HEAR THE DIFFERENCE.

Follow my reasoning:
Chances are, somebody, somewhere, under some conditions, can hear the difference. Does one single person hearing a difference satisfy your odd definition of "loss of quality"? It must, if you say it's entirely subjective.

So if I have a 192kbps VBR mp3 file that I clearly hear as being far less punchy in the bass than the original CD audio, while somebody else can't hear the difference on the same file, your argumentation ("indistinguishable to the original, hence no quality loss") would mean that my mp3 is of worse quality than his?

I don't think so.

The other thing you claim is that the psychoacoustic encoding is supposed to fool the brain into thinking we're hearing "CD quality". This is WRONG, for several reasons.

#1.) There is no such thing as a "CD quality" absolute. Every CD varies in sound quality, depending upon a multitude of factors, many of them located before the CD is ever pressed, and many more between the CD's data layer and your eardrums, and quite a few more located between your eardrums and your consciousness. There will be a clearly audible difference in sound quality between a run-of-the-mill CD of Britney Spears' latest album on an $80,000 stereo and a high-end original master gold edition CD of Steely Dan's "Gaucho", listened to on a 1989 Sony Discman while walking through traffic. Which, exactly, is "CD quality"? Also, why "CD quality"? Why not "2-inch analog tape quality?" or "Linn Sondek LP12 turntable vinyl quality"? Or "sitting in front of a chamber orchestra in a concert hall quality" (since no recording on any stereo system ever produced by man comes even *near* indistinguishably close to that)?

#2) Given that there is no such thing as a single, defined "CD quality": The psychoacoustic encoding is supposed to fool the brain into NOT NOTICING THAT SOMETHING IS MISSING. Actually the opposite of what you claim.


We can split words all day, but the question WRT lossy codecs is not whether ther IS a loss of quality, but only at what point it becomes acceptable to your listening situation/habits.

The end.

-s*

[Busemann]

My point is that as long as YOU don't hear a difference after doing an unbiased blind test, then there has been no loss of quality. I'm sure bats could hear a difference, but that's not the point.. The perceived quality can stay the same even after 9/10th of the data is removed

It seems like this turned into a discussion on semantics more than anything else

[bmedina]

Originally posted by analogika:
PERCEIVED QUALITY != QUALITY.

Lossy formats LOSE AUDIO QUALITY. Period.

Whether you can hear it or not.

I believe these statements are fundamentally misguided, so we'll have to agree to disagree.

BTW, if information == quality, how do lossless codecs throw away half the information without losing any quality?

[analogika]

Originally posted by bmedina:
I believe these statements are fundamentally misguided, so we'll have to agree to disagree.

BTW, if information == quality, how do lossless codecs throw away half the information without losing any quality?

WHa?

*That* is a fundamental misconception.

The whole point of LOSSLESS is that they DON'T "throw away" ANY information whatsoever.

Have you never received a .ZIP file or a .sit file of a Word document? They don't compress by REMOVING data, since that would mean that you'd end up with only, say, one in three characters of your document. Lossless compression works by reducing redundancy, by modulating data with certain algorithms that in turn create patterns which can be more easily reduced to simpler forms, etc. - there are numerous tricks employed, but NONE of them involve actually throwing out information. If they did, no .zipped, .tarred, or stuffed application or document could ever work once you decompressed it.

Lossy audio codecs, by contrast, employ psychoacoustic tricks based on the way we hear. The simplest example is a phenomenon called "masking": when two audible events take place simultaneously, the louder one will be the focus of our attention, thus "masking" the softer one. So a lossy codec will analyze the audio data for such masking incidents and simply discard the lesser events. The soft guitar strum under a cymbal hit isn't something that you can think back into the music, it is simply GONE. Chances are quite good that you're listening on earbuds, or in the background, so you may not notice the deterioration in quality, but that does not mean it isn't there. Masking is just one example of many, and while codec programmers are getting better and better, the fundamental problem is the same: What's gone is gone, and what goes is decided by some standards that directly influence what you hear.

The article by James Boyk above makes one thing perfectly clear: There is an awfully large aspect of hearing that doesn't actually enter consciousness. People hear sounds far in excess of 20 kHz (far beyond the commonly accepted hearing limit), AND IT AFFECTS THEIR BRAINS, even though they themselves profess to hear absolutely no difference.

Point of all the above is: the concept of "audio quality" applies completely REGARDLESS OF WHETHER YOU, PERSONALLY, THINK YOU CAN HEAR A DIFFERENCE.

See it as a Readers' Digest for audio. 9 out of 10 readers will never notice where exactly they're only getting a third of what Melville actually wrote, but there is no way you can argue that the condensed version will ever have the same quality as the original, even if they can't tell.

-s*

[bmedina]

You're still not understanding why your statement is incorrect, and I'm tired of explaining it to you. You've contradicted yourself in nearly every post.

The whole point of LOSSLESS is that they DON'T "throw away" ANY information whatsoever.

Contradiction. They do throw away redundant information that is not needed to reconstruct the original waveform. That's how they achieve compression.

They don't compress by REMOVING data....

Contradiction. It's really hard (read: impossible) to compress without removing data.

Lossless compression works by reducing redundancy

Yes, If the information is redundant, you can remove it without losing quality. This directly contradicts your earlier statement that information loss implies quality loss.

And I would avoid pointing to a single paper (which itself comes to no hard conclusions) to support your argument when a wealth of research contradicts it.

[analogika]

Originally posted by bmedina:
Yes, If the information is redundant, you can remove it without losing quality. This directly contradicts your earlier statement that information loss implies quality loss.

Suffice it to say that I consider "data loss" a situation where information is LOST.

A situation where redundant data is replaced by information on how to reconstruct it EXACTLY is NOT "data loss".

If this concept is not clear, then yes, we will continue to misunderstand each other.

Originally posted by bmedina:
And I would avoid pointing to a single paper (which itself comes to no hard conclusions) to support your argument when a wealth of research contradicts it.

There is NOTHING, NOTHING AT ALL that you could POSSIBLY post about lossy codecs that would contradict the research presented in that article.

Unless of course you really are implying that no information is lost at 192kbps AAC lossy compression.

In which case, you are completely and utterly wrong, and provably so.

The "wealth of research" you claim merely states that people are unable to CONSCIOUSLY HEAR a difference. I AM NOT DISPUTING THIS.

Everything else I can say, I have said at least three times.

I apologize for apparently not being able to make myself clear enough for you.

-s*

[frankiec]

Originally posted by bmedina:
Contradiction. It's really hard (read: impossible) to compress without removing data.

Wow, that's completely wrong. Are you sure you meant to write that?

For example, have you ever zipped a file? That's compression. Data is not lost.

[bmedina]

I'm talking about data as in bits on the disk. You guys are talking about data more abstractly, as in the data that defines a waveform.

Lossless compression is about removing unnecessary data, while preserving the ability to recover that data later. Nevertheless, in losslessly compressing a file from 50MB to 25MB, you've lost 25MB of data, even if you are later able to reconstruct the missing data. You seem to be stuck on the word "lost." Maybe "discarded" makes more sense to you?

As to high-frequency hearing (and specifically the article you linked), see here: http://www.hydrogenaudio.org/forums/...73&hl=boyk
Even if we could hear those high frequencies, most speakers are designed with a roll-off slightly above 20kHz.

[frankiec]

Originally posted by bmedina:
Contradiction. It's really hard (read: impossible) to compress without removing data.

What crack are you smoking? How come you won't answer this? What do you think ZIP, RAR, BZIP2, LZW, etc., etc., etc., are?

Originally posted by bmedina:
Lossless compression is about removing unnecessary data, while preserving the ability to recover that data later. Nevertheless, in losslessly compressing a file from 50MB to 25MB, you've lost 25MB of data, even if you are later able to reconstruct the missing data. You seem to be stuck on the word "lost." Maybe "discarded" makes more sense to you.

What crack are you smoking? How does a program remove "unnecessary data" then recovers "that data later"? That's impossible. A program can't read your mind.

Okay. I'll try to make this as simple as possible to show you that lossless compression is NOT removing/losing/discarding any data.

Fact: lossy: Refers to data compression techniques in which some amount of data is lost.

Fact: lossless: File compression that involves no loss in data and therefore no loss in quality.

---

WAV encoded to MP3 encoded to WAV

$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5
$ md5 song.mp3
MD5 (song.mp3) = e3e4a22598fcfd11c8237b0f35b642bb
$ md5 song.wav
MD5 (song.wav) = bdc3a8486bc3700fce0085f318db20d1 !!!

Conclusion: MP3 is LOSSY. Data is NOT COMPRESSED. DATA is REMOVED/LOST/GONE FOREVER.

---

WAV encoded to AAC encoded to WAV

$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5
$ md5 song.m4a
MD5 (song.m4a) = 3ca272d744a34127352ae31ef1182855
$ md5 song.wav
MD5 (song.wav) = 27d99802e236b7e9b47b332fefff4d8a !!!

Conclusion: AAC is LOSSY. Data is NOT COMPRESSED. DATA is REMOVED/LOST/GONE FOREVER.

---

WAV encoded to Apple Lossless encoded to WAV

$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5
$ md5 song.m4a
MD5 (song.m4a) = 429fb794e844d542ef51b148c240a2c8
$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5

Conclusion: Apple Lossless is a LOSSLESS format. No data is removed/lost/thrown away.

---

WAV compressed to ZIP uncompressed to WAV

$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5
$ md5 song.zip
MD5 (song.zip) = 6adaa43e1d4de337b661015070bc38ef
$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5

Conclusion: ZIP is a LOSELESS format. No data is removed/lost/thrown away.

---

Thank you, thank you, thank you all. You can stop the applause now.

[bmedina]

Did you even read my initial post?

Lossless means that the decoded output of a file will be bit for bit identical to the input.

The codec compresses, discarding data (notice I've stopped using the word "lost," since it apparently causes confusion), then it reconstructs the discarded data to produce an output that is identical to the initial input.

Are we all clear yet?

WAV encoded to Apple Lossless encoded to WAV

$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5
$ md5 song.m4a
MD5 (song.m4a) = 429fb794e844d542ef51b148c240a2c8
$ md5 song.wav
MD5 (song.wav) = 21342ad35115bc2a4b8c0f852bcafcb5

Conclusion: Apple Lossless is a LOSSLESS format. No data is removed/lost/thrown away.

You may have noticed that the intermediate file is smaller than the initial file. That's because the codec has discarded data that is not necessary to reproduce the initial file perfectly.

The more interesting discussion of whether a change in the waveform implies quality loss has been lost in this bickering over semantics. Perhaps we can move on?

[frankiec]

Originally posted by bmedina:
Did you even read my initial post?

The codec compresses, discarding data (notice I've stopped using the word "lost," since it apparently causes confusion), then it reconstructs the discarded data to produce an output that is identical to the initial input.

OH MY ****ING GOD. You're a ****ing moron. I just showed your the md5 hash results. Programs can't create data from thin air. The data is compressed. Whether it's Apple Lossless, ZIP, RAR, or whatever. THE DATA IS NOT DISCARDED. IT IS COMPRESSED. (RAR is exceptionally good at dictionary methods. Do you know what that is?)

Originally posted by bmedina:
You may have noticed that the intermediate file is smaller than the initial file. That's because the codec has discarded data that is not necessary to reproduce the initial file perfectly.

Um, no. The data is compressed, moron. Go find out how compression works.

Repeat: Programs can't create data from thin air.

Repeat: Programs can't create data from thin air.

Repeat: Programs can't create data from thin air.

Go away. I'm not discussing this with you anymore, you're a ****ing moron.

[frankiec]

Originally posted by bmedina:
The more interesting discussion of whether a change in the waveform implies quality loss has been lost in this bickering over semantics. Perhaps we can move on?

By the way, moron. There's no change in the waveform from WAV/AIFF to LOSSLESS to WAV/AIFF...

... The waveform is a visual representation of the data.

*sigh*

Moron.

[analogika]

Originally posted by bmedina:
Did you even read my initial post?

The codec compresses, discarding data (notice I've stopped using the word "lost," since it apparently causes confusion), then it reconstructs the discarded data to produce an output that is identical to the initial input.

Are we all clear yet?

If you are going to insist on continuing to argue semantics, it really doesn't help if you simple USE THE WRONG TERMS.

"Audio quality", when talking about source information, is NOT DEFINED BY WHAT YOU CAN HEAR, but BY WHAT IS THERE.

Please read my above half-dozen posts.

If you say audio quality is entirely subjective, then that MUST mean that the exact same file is of different quality if you can not hear the difference, but I can.

That is a completely brain-dead assertion.

I'm done with this.

-s*

[bmedina]

The level of reading comprehension around here reminds me why I post so rarely. If you guys are going to resort to personal insults, then yes, the discussion is over.

I'm not sure why people are allowed to make posts like frankiec's without moderator intervention.

[analogika]

While that is true, it won't your posts valid.

[lenox]

frankiec needs a firm kick to the pants.

back on-topic somewhat. aac vbr encoder? i thought aac was vbr by default...can someone point to a link to read so that I could educate myself here?

[Busemann]

Originally posted by lenox:
back on-topic somewhat. aac vbr encoder? i thought aac was vbr by default...can someone point to a link to read so that I could educate myself here?

It's not real VBR, but it uses a technique called bit-reservoir, which means the bit-rate can vary somewhat depending on the complexity of the music. It doesn't use bits on silence either, unlike the iTunes VBR mp3 encoder(!)

A proper AAC VBR encoder is included in QuickTime 7.

[lenox]

Interesting! Thanks for clearing that up.

[gzeus]

Originally Posted by Busemann

A proper AAC VBR encoder is included in QuickTime 7.

Is this a definite statement? If it is, it would be the first mention of it on the internet. I currently use EAC for secure rips and Nero aac vbr for the encoder, but I want to ditch the pc and use xAct or cdda2wav for secure ripping and itunes vbr aac for the encoding.

[bmedina]

You haven't read the entire internet, then. The VBR AAC encoder in Quicktime 7 has been known for some time (at least since WWDC 2004).




However, there is no HE-AAC encoder.

[sehested]

@bmedina:
What an outstanding informative post you made to the original poster. Excellent.

@analogika:
I don't generally like people shouting in their mails, but you have made quite a few good points too.

Ignoring the risk of beeing flamed I will stick my neck out in attempt to define audio quality

Audiophiles in general are very concerned about the fidelity of the reproduced audio and go to great lengths to achieve the best possible absolute quality.

To name a few things along the process:
- special recording techniques to minimise the number of amplifier steps between the recording and mastering. Pop music mixing typically involve the use of 100s of amplifiers, each modifying the signal slightly. Special connaiseur recordings use as few as three amplifiers between the recording and mastering in order to preserve as much as possible of the original depth of image, timbre and dynamics.
- use of special media, such as heavy vinyl, SACD or DVD-A
- use of no compromise components in stereo
- use of no compromise components in speakers
- specially build listening rooms sometimes costing several hundred thousand dollars

Still after making all these steps to ensure fidelity and highest possible absolute audio quality audiophiles agree that nothing beats being present at a live performance.

Another, less talked about thing, is that the majority of audiophiles will fail to hear the difference between their expensive equipment and more modest budget alternatives when submitted to double blind tests.

The reason they fail the blind test is because what they hear is the perceived quality, not the absolute quality.

However you will be able to find audiophiles that can tell the difference. Some even have golden ears and are paid good money by speaker manufacturers to test their products.

So much for the audiophiles.

When it comes to lossy formats focus changes from the absolute quality to the perceived quality.

Radio broadcasters have invested a lot of money in trying to find the sweet spot for their satellite radio broadcasts.

They have been conducting organised listening tests to find out how people perceived the audio quality at different bitrates.

Remember now, some people will always be able to tell the difference. So for each lossy format at each bitrate they listed:
- percentage of people not being able to tell a difference
- percentage of people finding the difference slightly annoying
- percentage of people finding the difference annoying
- percentage of people finding the difference bad
- percentage of people finding the difference unbearable

The sweet spot was then set at the bitrate where 95% of the listeners where unable to tell the difference.

At the sweet spot 5% of the listeners are still able to tell the difference and some will even find the difference annoying or worse.

However whether you personally are amoung the 5% or the 95% nobody can answer but you.

At the center of the continued development of lossy formats is this concept of perceived quality.

It is so central that the word perceived is often left out - the same way that audiophiles leaves out the word absolute when they talk about quality.

Whenever these two quality definitions clash you will see a religious war emerging - taking over any sensible argument.

This being a thread discussing lossy formats and not audiophile equipment, implies that the word quality, whenever mentioned, is refering to the perceived quality and not the absolute quality.


To finish off, here's a teaser for you all:
An inventer has discovered that hypersonic sound transmitted directly towards the human brain will be perceived by that human as audible sound!
Soundless sound system

[analogika]

Thanks for the precise and comprehensive summary of this thread.

Yes, I'm from a music *creation* background, and to me, the whole argument was centered around clarifying the difference between absolute sound quality (fidelity) and perceived sound quality (i.e., the "good enough" factor).



n.b.: in sound creation (studio work), the various different amps and processors are usually employed precisely *because* they affect the sound. Remember, this is about sound *creation*. The point of using a Fender stack on an electric guitar, or sending vocals through the weird old tube-amp that the sound engineer found in his father's attic, is that they AREN'T true to the original signal. This is a desired effect.

That is a completely different situation from what happens during playback of the final product.