HDX rip better than iTunes rip?

quote:

Originally posted by Harry H. Wombat:
@matt303

The claim has been made, by NAIM or NAIM guys, that the HDX rips sound better than iTunes and this claim is also made in HiFi Critic.

And I'd say it again. Do the comparison on your own, as you don't have to take my word for it. EAC rips ALSO sound better than iTunes rips, when configured correctly.

Ive listened to an HDX against a M/L combo and the HDX wasn't any better, in fact the M/L was to my ears better.

What this reflects is what harry clearly refers to, 'subjectively'

This thread was linked to an article which the author, a sound engineer tested under Scientific conditions different rip software. He found that there was no difference between the software, they all ripped 'bit for bit' this included EAC & iTunes.

So, EAC & iTunes rip bit for bit FACT.

What people are making statements about is their own subjective interpretation, which is fine, hence most of us prefer the Naim sound to other HIFI.

But don't pass of your own subjective interpretation as FACT, it isn't!!

Basically any difference heard between identical files on the same hardware is completely down to your perception being altered by conscious and subconscious expectations. A properly conducted blind test is the only way to check if identical data played through the same player sounds the same. Any other type of test simply proves that "believing is hearing" which is a factor in any none blind test.

@gary1 & David

I repeatedly make the same statement because it is scientific fact. I have no problem intellectually with an HDX rip sounding better than an iTunes rip (to be open and honest I do not believe that it sounds better but I cannot rule out the possibility that it does).

I also have no problems having my tests (which showed and XLD rip to be bit-wise identical to an iTunes rip where XLD is claimed to be superior) to be shown incorrect. No problem at all.

My problem is that people who should or do have the capability to check the files do not do so and do not think it is necessary to do so. In another place, people who believe that bit-identical files cannot sound different are described as "foolish". Now - this really gets me firing on all cylinders

This argument is firmly within my discipline, that of computing. Ripping and replay (until the signal enters a DAC) is nothing at all to do with Hi Fi - it is a pure computing problem to do with data harvesting, transfer and sometimes functional transformation. All subjects that are more than well-understood at the theoretical, practical and pragmatic level. After the signal enters the DAC I am out of my depth and fully admit it. Some of the brightest people who have ever lived have been thinking about and designing this stuff. The statement which I make constantly that two bit-wise identical files replayed through exactly the same chain have to sound exactly the same is a mathematical fact. By denying this (or more accurately denying the need for testing) casts doubt on the whole of mathematics, computing and the basic structure of the universe. Really - I haven't gone over the top there. In this case,would it not be at least respectful to test the files?

I admit that I have not done the subjective tests that you have but I have no immediate access to an HDX neither am I in the market for one. I don't feel comfortable visiting a dealer and taking up their time when I shall not be buying. (Actually I am buying a NAIT for the kitchen so maybe I can use that as an in).

The issues with subjective tests are well-known and any manner of subtle "signals" sent or received from the environment can change one's perceptions of what one is hearing. Indeed, many times a single track played consecutively will reveal different things each time which one should welcome as a joy!

At the limit this becomes a subjective/faith based argument against an objective/fact based argument and these never terminate in agreement The s/f side hears a difference and therefore sees little point in testing. The otehr side tests and see little point in listening. It is a gap that cannot be closed.

I know you guys do not agree with me but thanks for keeping the discussion polite and I endeavor to do the same from my side.

I remember similar objectivist arguments made when discussing differences between CD pressings-the storage and data objectivists claiming that there should be no difference in sound.

Again, and in all fairness, when you ask the people who design these storage subsystems at an electronic level (say, storage ASICs), you get a different take on the process (especially when you understand the degree to which CD data and CD-ROM storage differ from a heuristic perspective).

@David

I understand the point which if I may summarize in my own terms is that the objective argument requires you know everything there is to know and even if you think you know everything there still may be things we do joy yet know. The unknown unknowns as has been said in another context.

In this case, though, we posit the replay mechanism is exactly the same, the media is exactly the same and the only thing that differs are the files. Now if we posit that the files are exactly the same then there are no gaps for unknowns and the sound must be the same. There is no way out.

Now if the files sound different then given same media and replay the only difference in the system are the files. Therefore the conclusion is the files are different. And this is an easy observation to make only the people making the claims choose not to observe.

I think I can go no further with this or offer better explanations of the position. I am also more than happy for anyone to point out flaws in the logic. I am always happy to be wrong. So the base position which I will not restate is as above and I'm done with it.

[EDIT: There would be a gap in the logic if there were such a thing as a NAIM shaped bit or an iTunes shaped bit. There is not. Writing the bits is the responsibility of the drive - if it is told to write a "1" for instance it writes a "1". There are also no "gaps" between the bits in which a little devil might reside.]

Perhaps we need to clarify a few points, in a relatively simple manner (and I am grossly oversimplifying this):

Every data storage system, by definition, is an analogue process (inductive pickup head, in the case of a Winchester hard disk drive) and, as such, has some degree of data corruption due to transfer of energy or state.

There exists a smallest "cell" of data, if you will, which defines the granularity of the data storage system.

There exists a means to organize these cells into addressable locations. These locations then can be grouped into multiple levels of organization. A scheme can then be developed to manage these addressable levels.

Energy or state transfer is, by definition, a lossy process, which can be quantified by a hysteresis curve. As these are analogue transfer functions, they can be multiplied to show aggregate energy transfer loss.

There is a pair of thresholds that define the transfer from analogue storage to two-state binary digital storage.

If the aggregate energy transfer loss for a specific storage cell biases the stored analogue value to fall between the values that define the two binary states, the cell can no longer be trusted to store and retrieve specific analogue values that match their mapped binary values.

As this can be measured on a per-cell basis, it can be mapped within the storage management schema.

A storage management schema cannot be self-contained within the storage itself (as the integrity of the data itself cannot be determined).

Error correction transfer functions can be classified into two types: homomorphic and polymorphic.

Homomorphic error correction is by definition a one-to-one mapping; the amount of space required to describe a binary homomorphism IN BINARY is equivalent to the cardinality of the number of storage cells. A system such as this requires similar overhead to that required for actual data storage.

The overhead for storage addressing and integrity management (e.g., an inspector within an underground storage vault) can be heuristically tallied and averaged for typical, large data transfer operations, but is still subject to the requirements of individual organized levels for data retrieval, subject to the integrity of the mapping (in the case of a Winchester hard disk drive, of stored magnetic analogue values).

Now for the jump forward...

Disc parity schemes for Winchester hard disk drives based on odd, even or double parity are polymorphic–they cannot be used to reconstruct the data for a specific given organized level, as there are more than one data set that maps to that parity value.

and...

No two hard disk drive platters have exactly identical storage capacities at the cellular level.

so...

No two data files will ever be stored in identical locations or with identical induced analogue values at the hard disk platter level. (This is one of the reasons why, for example, disk access time varies by storage location–just as it does when accessing printed or microfiched data in an underground vault.)

The success of this is dependent upon the drive control electronics achieving the following things:

- correctly mapping analogue values from the pickup head to their two-state binary equivalents (relatively easy in a magnetic storage system, but subject to various electrical wear factors at the pickup head, as well as circuit-speed issues)

- adequately mapping storage at a cellular level to the proper organized levels while mapping out "bad" cells or levels

- providing a schema for error reporting at the appropriate organized level

- filling the buffer with data in a sufficient manner to achieve ideal drive data transfer bus speeds against the heuristic constraints of error correction, seek time and data integrity subject to management overhead

So it is impossible to speak of "identical" files stored on hard disk drives without speaking of other differences, or, on the other hand, ruling out their parametric effects as inconsequential to the discussion.

The degree to which these effects are inconsequential depends on the area of focus when describing the storage system.

And I haven't even hit optical disc or FLASH-memory storage yet....we cannot make assumptions at the file level that two files are identical unless the process use to store and retrieve the data is error-free.

Which it isn't–as every storage medium is analogue, and, therefore, lossy due to the same factors that affect every other physical process.

@David

Relatively simple

Although I do not believe this defeats the original hypothesis as it was intended. If I was to insert a clause such as "the files will either sound identical or if different then different in a non-deterministic manner (no one ripping mechanism can be shown to be superior)"

This is starting to feel like ever decreasing circles though. The first person to mention quantum mechanics in support of the HDX ripping mechanism gets a slap

I've had enough as I suspect have many others on both side. The two sides will never meet as there is no middle. But one should not dismiss the underlying importance of this thread and what it means outside of the narrow confines of this discussion.

The thread may, to many, have become boring but do not dismiss lightly what it is talking about.

It is interesting (and somewhat educational) to know that, at the cellular level, stored data is in an analogue form and has to undergo a potentially lossy transfer process to binary form before it is processed further.

However, IMO this is again irrelevant to the discussion at hand since any errors in the (upstream) analogue to digital transformation would show up as differences in the bit by bit comparison between two files. If there are no differences revealed by the bit level comparison, the data being fed to the DAC stage (from the two files that are being compared) are identical and, as stated above, the resulting sound should also be identical.

Must say also that (as mentioned above) it is a sad day indeed when some prestigious and respected hi-fi companies are resorting to pseudo-science and general BS to sell (some of) their products. Better to focus on genuine innovation and engineering excellence to do this, I would think, but ...

With all due respect, jazzfan is missing the point-there are no bit-for-bit comparisons without access to the original 16-bit dithered file (in the case of Compact Disc), which still gets interleaved into its CD-DA storage form, before being "ripped" and re-assembled into PCM samples that resembles (mostly) the original file.

How "mostly", if you will, that data on the distributed media first resembles and secondly is de-interleaved (with PQ subcode removed) and error-corrected to present itself as data is the point of this discussion. There will be errors, and they will need to be corrected. (In the case of discs with Midbar or other copy-protection systems, we know this to be a fact.)

Taking the Compact Disc out of the provenance of delivery eliminates the variable that we are discussing here-but make no mistake, there seems to be perceived differences that require some explanation of what occurs as the CD-DA data is re-constructed, to whatever degree of attempts at integrity is provided.

And then-it must sound good.

quote:

Taking the Compact Disc out of the provenance of delivery eliminates the variable that we are discussing here...

My impression is that this thread is about differences between the digital files produced by different ripping software apps and, if so, that would seem to eliminate the variable you have raised from the discussion, once the data in those ripped files are identical (as was demonstrated by the experiments referred to by the OP).

David... very interesting. Thanks for your efforts.

To the OP....

I have read through this thread, for some reason, and it finally dawned on my to look at the thread's Title.

The answer is YES, anything is better than an iTunes rip! In my research I would only want to rip with EAC or some very similar (HDX Engine).

If I owned an HDX I would be a happy cat. But probably still do my rips on a PC. I like FLAC, tags, etc....

I require the following which I do not believe iTunes can accomplish.

- OFFSET CORRECTION!!!! very important (HDX has work-around by using drive with zero offset)
- Pre-gap tracks (hidden information)
- FLAC
- Log File
- Cue File
- Detection of pre-emphasis (rare I know, but I have some and looking for others Toto IV...Japan)

I am not going to get caught up in "bit perfect", and "sounds better". My hearing threshold, while large in some areas, completely fails me here. I cannot hear a difference between an iTunes WAV rip and an EAC WAV rip, and I imagine an HDX rip. All of which are said to be "bit perfect". Which I dont understand in the case of iTunes. How could it be?

BUT, the EAC rip (and HDX)is surely "better" as it contains ALL of the information on the disc. ALL. Period.

Since it is obviously pointless and bordering on quantum logic to discuss "bit-perfect", all I strive to do is be able to rip and playback a CD and have the experience be identical to playing the CD itself.

If I could perfectly synchronize the start of the playback of a CD and its ripped counterpart, I would expect them to play identically. Perfect match in timing, exact length of gaps between songs, any information (music) before track 1 (exists on thousands of CDs). I should be able to flip back and forth and not even know which one I am hearing.

Of course I am only talking about "information" here. Of course they will "sound" different, even if going through the same DAC.


My point is, rip with EAC if you have a PC, rip with an HDX if your fortunate to have one of those. If you have a Mac, use XLD. iTunes is a store, not a ripper. You might as well just use Windows Media Player..... at least you could use FLAC then

WAV from iTunes even sounds better than Apple Lossless, just as there's a sonic difference between levels of FLAC compression.

dont forget this classic



or this...

Regarding off-set levels.

Spurred on by a suggestion on this Forum (on of the few interesting ones on this subject) - that offset may affect the boundaries of the data blocks as they enter the DAC - with a resultant an audible affect - a few weeks back I used XLD to rip Thea Gilmore's - Lipstick Conspiracy (NAIM label CD).

Several times - each rip with a different off-set; making sure the off-set auto-correction mode was disabled. Some of the offset ranges I used were so large - that yes - I COULD tell a marked difference in the rip - the rip actually being of the wrong album track (quite confusing). For the range off-sets that resulted in ripping of the target track - the size of the resultant rip files were indeed different - as one might would expect.

However - played through my System - I could detect no audible difference.

If you have a MAC/DAC combination a quick / easy experiment that I'd encourage to you to try and report your results.

A final comment. As per Harry's posts.

IF two copies of data are identical on disk (however each was produced is immaterial) - the sound will be identical when played through the same audio chain.

I'm assuming there is general consensus on this point and that no one is arguing the contrary? If one were to argue the contrary - the implication would be not no ripping system (including the HDX) could EVER achieve consistent ripping results.

Think about it

Richard

David, that is all very interesting, but if we take Kent Poon seriously it is all compeletely irrelevant. The only way the physics of the drive matter is if the analogue states on the drive are not read correctly into the ones and zeros--and that is a very rare occurrence indeed.

According to Poon, there is no way to tell his ten different rips (with different drives and software engines) apart whatsoever, whether by checksum or by looking at the actual waveforms reconstructed from the data. If you line up the iTunes rip with Wavelab or whatever, it is exactly the same.

Furthermore, he shows that the output from itunes as a player is the same as his other software (including his pro stuff). [Note this is on a Mac.]

The onus is clearly on those who claim a difference to show a difference. This is not like arguments over the audibility of different amplifiers or even stands and wires, where physical differences are objective but the debate is over audibility. Here we must first establish that there is even a phenomena to test subjectively.

Like, take digital output of an HDX rip and compare it to an itunes or EAC rip.

Or put several different rips on an HDX and then measure the ANALOG output when playing back. Dump it to digital and compare waveforms. What do we see?

Now I might be a silly audiophile, but I believe if you dumped the output of my CDS to hi-rez digital and analyzed waveforms, it would look different than the output of my 18 year old Rotel. Arny Kruger et. al. could then claim it was of no consequence, but there would be something to argue over.

However, I am not at all sure you would see identifiable differences in the analog out when playing an Itunes rip and an HDX rip on an HDX.

quote:

Originally posted by Eric Barry:
David, that is all very interesting, but if we take Kent Poon seriously it is all compeletely irrelevant. The only way the physics of the drive matter is if the analogue states on the drive are not read correctly into the ones and zeros--and that is a very rare occurrence indeed.

According to Poon, there is no way to tell his ten different rips (with different drives and software engines) apart whatsoever, whether by checksum or by looking at the actual waveforms reconstructed from the data. If you line up the iTunes rip with Wavelab or whatever, it is exactly the same.

Furthermore, he shows that the output from itunes as a player is the same as his other software (including his pro stuff). [Note this is on a Mac.]

The onus is clearly on those who claim a difference to show a difference. This is not like arguments over the audibility of different amplifiers or even stands and wires, where physical differences are objective but the debate is over audibility. Here we must first establish that there is even a phenomena to test subjectively........

Agree completely - exactly my take on the matter. Especially the first and last paragraphs.

Eric,

Agree completely, great post.

quote:

Originally posted by paremus:
For the range off-sets that resulted in ripping of the target track - the size of the resultant rip files were indeed different - as one might would expect.

However - played through my System - I could detect no audible difference.

IF two copies of data are identical on disk (however each was produced is immaterial) - the sound will be identical when played through the same audio chain.

Aren't there a few missing 'bits' of logic here?

Presumably your two (admittedly at extremes of viability) rips were different (on disk) - and yet sound indistinguishable to you.

Does ANYTHING sound different?

There are no missing bits of logic.

All I know is I tried to rip 2 CDs last night in iTunes (on a Mac). Both CDs have pregap information before Track 1. One of them has a three note lead-up that segues directly into the 1st track. The other (ambient) has a fade-in that leads into the first official track.

NEITHER were detected and ripped with iTunes.

So whether or not the parts iTunes DID rip were bit perfect is a moot point. The iTunes rip didn't capture the whole album.... therefore not bit perfect according to my definition.

Harry, iilyama, jazzfan etc,
I am not sure what your points are anymore. Is it that EAC rips as well as the HDX? IF so you are right.

Is it that iTunes is also as "good"? Well, in that case I feel you are wrong.

If you cannot hear a difference between all of the rips, why get caught up in the discussion?

Clearly there is more going on than meets the eye. Perhaps there are a bunch of placebo effects in conjuction with Emperor's Clothes.

In the end, if an HDX/EAC/iTunes user "believes" their rip sounds better, more power to them....

@pc

The points are as they have always been. I do agree with you, though, on the points that you make.

If you cannot hear a difference between all of the rips, why get caught up in the discussion?

Because I know my subject and when I see bad computing science I just cannot let it go. But maybe the challenging has gone on long enough. The points are made and they are here for anyone who wants to find them.

Harry,

Since this isn't a Computer Science Forum, what exactly are you trying to achieve?

What do you want "us" to learn with your comments?

What should "we" be doing differently?

I see that you understand parts of this subject very well. What I am lost on is your motivation. Not being pejorative in the least. Seriously, what am I supposed to glean from this thread?

If you think some are simply wrong about how the 0s and 1s work, I get it.

If it is something else you are getting at, I would be interested in specifically what that is.