[limpidglitch]

Sure, but the issue then is with the loudness war, and not Apple. To Apple this is a just a precaution, they don't want to sell damaged goods.
If anything, Apple is actively against the whole loudness hysteria. At least going by how they manage playback levels through their radio service using perceived loudness, à la BS1770.

 

[bigshot]

   
The louder master should never have been made either, so you are not comparing apples with apples. A properly mastered 16/44 will convert just fine to AAC. Any "mastered for iTunes" label is nothing more than marketing nonsense.

Encoding using iTunes, I've had tracks that sounded just fine on CD clip when I did a straight conversion to AAC. I just lowered the level overall a few dB before converting and it encoded fine. AAC tends to boost the level a bit as it compresses.

 

[Greenears]

Sure, but the issue then is with the loudness war, and not Apple. To Apple this is a just a precaution, they don't want to sell damaged goods.

If anything, Apple is actively against the whole loudness hysteria. At least going by how they manage playback levels through their radio service using perceived loudness, à la BS1770.

I don't have a problem with the concept. But then why doesn't Mr. Stewart just provide some solid technical information. It seems like we are all just left guessing.

 

[limpidglitch]

He's a business man, förstås, selling a service.

 

[Greenears]

He's a business man, förstås, selling a service.

Well there are more than one way to sell a service.

 

[efeuvete]

I think I have something to say here. This:
Before any noise proccesing (Your text is very clear on this issue) the system must "take" samples of an analogic audio signal; And it is clear that the error in taking samples in 24 bits is 256 (2^8) times lesser than in 16 bits. In other words, 16 bits has a "minimun fixed" error of about 15,6 parts in a million and 24 bits only about 0,06 parts in a million.
So, the point to me is: Can my ears notice a 15,6 part in a million error in every sample of every sounding sound wave?
I woud say that, listening closely and carefully, yes, they can (And maybe more the higher the frecuency, which is logic because the higher the frecuency, the less number of samples per wave).
Then, it is clear to me that, even not being sure of guessing a "16/24 blind test" in any sound situation, I'm sure I'm going to have a better experience all along 5 minutes of any music in 24bits than I'm going to have it in16 bits.
And anyhow, last but not least,... God bless mp3, 128 Kbs for quite a good bunch amount of reasons!

 

[headwhacker]

^ I hope you read the thread from start before you post that comment. Nevertheless, welcome to head-fi

 

[RRod]

  I think I have something to say here. This:
Before any noise proccesing (Your text is very clear on this issue) the system must "take" samples of an analogic audio signal; And it is clear that the error in taking samples in 24 bits is 256 (2^8) times lesser than in 16 bits. In other words, 16 bits has a "minimun fixed" error of about 15,6 parts in a million and 24 bits only about 0,06 parts in a million.
So, the point to me is: Can my ears notice a 15,6 part in a million error in every sample of every sounding sound wave?
I woud say that, listening closely and carefully, yes, they can (And maybe more the higher the frecuency, which is logic because the higher the frecuency, the less number of samples per wave).

 
Everyone talks about extra levels from 24bits but doesn't bother to understand how they manifest themselves. The signal you get after picking a "level" is the original, un-rounded signal plus the rounding error that you get when you pick the level. At high amplitudes, that rounding error basically sounds like white noise. It's only when you get down to low signal amplitudes that the rounding error begins to manifest as harmonic distortion. The level of these errors is of course lower at 24bits, but that just results in a lower noise floor, and there is little music that gets anywhere the noise floor of even 16bits.
 
What you can hear doesn't matter if you're letting knowledge of the files bias you.

 

[cjl]

  I think I have something to say here. This:
Before any noise proccesing (Your text is very clear on this issue) the system must "take" samples of an analogic audio signal; And it is clear that the error in taking samples in 24 bits is 256 (2^8) times lesser than in 16 bits. In other words, 16 bits has a "minimun fixed" error of about 15,6 parts in a million and 24 bits only about 0,06 parts in a million.
So, the point to me is: Can my ears notice a 15,6 part in a million error in every sample of every sounding sound wave?
I woud say that, listening closely and carefully, yes, they can (And maybe more the higher the frecuency, which is logic because the higher the frecuency, the less number of samples per wave).
Then, it is clear to me that, even not being sure of guessing a "16/24 blind test" in any sound situation, I'm sure I'm going to have a better experience all along 5 minutes of any music in 24bits than I'm going to have it in16 bits.
And anyhow, last but not least,... God bless mp3, 128 Kbs for quite a good bunch amount of reasons!

Here's the thing: 15.6 parts per million works out to be (if properly dithered) a broad-spectrum noise floor (much like tape hiss) at about -110dBFS or so. That's not even close to audible unless you have the volume up so loud that a normal signal level would blow out your eardrums or your transducers (or both). Also, your logic about higher frequencies being affected worse shows why you should learn the math before just trying to guess about things like this - quantization error due to bit depth is frequency independent, and affects all frequencies equally (though you usually do shape the dither to have a higher noise level at high frequencies, simply because that is less audible).
 
In addition, a blind test has no particular time limits. If you have a "better experience" over long periods of time, that is testable. Fast-switch, level matched tests have been the ones people are best able to discern small differences on in the past though, so chances are that over 5 minutes, you wouldn't even notice fairly large changes, much less subtle ones.

 

[efeuvete]

My only intention posting here has been trying to clarify a bit ideas regarding 24/16 bits differences, not arguing just for the sake of arguing; I mean, maybe 15,6 parts in a million can have a different impact on differents ears or even no impact at all (Though my "bias" does not accept this last option) but... 15,6 parts in a million rounding error are 15,6 parts in a million regardless:
- How this rounding error is supossed to be manifesting (I mean, a complete wave with a 15,6 ppm rounding error to start from, in every one of its samples, does not make a distorted wave but a subtly different one).
- The volume of your amp, because what math says here is that there is a 15,6 part in a million rounding error, being the wave as big as a thunder or as little as a butterfly flapping.

 

[bigshot]

  I mean, maybe 15,6 parts in a million can have a different impact on differents ears

 
And if the wind is blowing just the right direction, a pig might fly.
 
The truth is, there are plenty of controlled listening tests on this and to date, no one has been able to hear a -110dB noise floor when listening to music at normal volume levels. So there is no reason to believe that there are ears out there in the wild that perform any better than the plethora of ears that have already been tested.

 

[efeuvete]

Bigshot..."no one has been able to hear a -110dB noise floor when listening to music at normal volume levels"...
It's you who talk about noise, not me. I'm talking about two different wave forms. I mean, Is a male voice a distorted female voice or viceversa?

 

[bigshot]

Do you think there is a human on earth that can hear distortion at -110dB?
 
It's fine to entertain yourself with interesting math problems, but we are talking about real world sound here, not just theory. There are hundreds of listening tests of redbook vs high bit/sample rate audio. They all come out the same.

 

[efeuvete]

I insist: I'm not talking about discerning distorsion nor noise against clear sound, but about different wave forms. An Eflat 5th octave note on a recorder and that very same note on a metal flute are just different, none of them is more distorted than the other.

 

[StanD]

  I insist: I'm not talking about discerning distorsion nor noise against clear sound, but about different wave forms. An Eflat 5th octave note on a recorder and that very same note on a flute are just different, none of them is more distorted than the other.

Whatever change to the waveform you propose would at best be too small for a human to discern. We are not comparing the timbre of flutes and recorders, that is so out of context and is not going to stick to the wall,especially at the 5^th octave.