AIFF for audiophile iTunes library?

Discussion in 'Digital Audio' started by 50548, Jul 20, 2013.

  1. Guest

    Joined:
    Apr 17, 2005
    Location:
    Currently in Switzerland
    #1
    Having recently invested in hi-fi/bit-perfect audio and considering that most of my library is already in ALAC format, I was wondering if the next major quality step would be to move it to AIFF format.

    I am asking this because a recently-read hi-fi publication stated that even ALAC would show quality differences when compared to AIFF, since the latter would NOT require PCM decompression upon reproduction and would still retain metadata compatibility with iTunes.

    Does anyone have thoughts or suggestions on this?
     
  2. macrumors 68020

    Prodo123

    Joined:
    Nov 18, 2010
    #2
    I used to be on the same boat. I even made a huge, somewhat embarrassing post about it a couple years back.
    I can confidently say that there is no difference in quality between any two lossless format, compressed or uncompressed. That is because the decompression necessary for playback can be performed MUCH faster than the sample rate of any recording out there (to give an example, a 2020MB ZIP file can be decompressed in 1 second, equivalent of 1,310kHz in 16 bit audio terms. Highest sampling rate used for audio mastering regular CDs is merely 192kHz). That means the decompression NEVER lags behind the actual audio, and the decompression buffer is never overfilled. There are thus no skips nor decompression artifacts present.

    You can even try this yourself. Try converting one of your ALAC files to AIFF, and time it. Then compare the time it takes to decompress it to the actual length of the song.
     
  3. thread starter Guest

    Joined:
    Apr 17, 2005
    Location:
    Currently in Switzerland
    #3
    Tks, Prodo - please see below the quote I am talking about, from an Audioquest white paper:

    "Since the original music file is restored bit-for-bit Apple Lossless files offer much better sound than lossy MP3s and is compatible with high-resolution music files. Interestingly, uncompressed music files such as WAV or AIFF can sound better than lossless compression formats like Apple Lossless or FLAC. Perhaps this is because they don’t require the additional step of being “unzipped” and restored to their original PCM data package during real-time during playback. Listen and you’ll hear the difference."

    As for your remarks above, I am still not sure whether there would be no difference between both formats - the link below contains a good discussion on that:

    http://www.computeraudiophile.com/f8-general-forum/aiff-or-apple-lossless-13050/index2.html
     
  4. macrumors 603

    Joined:
    Jun 30, 2007
    Location:
    Atlanta
    #4
    A BIG sack of link bait poppycock coming from a company that sells >$2000 HDMI Bull S.... cables.

    The day you see numbers different in a Zip spreadsheet will be the day you can find ANY difference in an ALAC (or FLAC) from the LPCM source.
     
  5. macrumors G4

    Joined:
    Jan 5, 2006
    Location:
    Redondo Beach, California
    #5
    There is absolutely ZERO DIFFERENCE. The lossless formats are in fact lossless.

    All you would do is waste disk space.

    ----------


    This is an centuries OLD trick that snake oil salesmen use. They say ONE thing that is true and then immediately make a totally unsupported claim. As if being close to a true statement adds credibility.

    So first he says bit-for-bit is better than MP3. Correct. But next he says "Interestingly, uncompressed music files such as WAV or AIFF can sound better than lossless " which is a complete fabrication that he made up with zero supporting evidence.

    He then continues assuming his fabrication is true.

    This is total BS and likely the author knows it. I'd bet later on he tries to sell something to you.
     
  6. macrumors G4

    Joined:
    Jan 5, 2006
    Location:
    Redondo Beach, California
    #6
    A better test is an "ABX" type test. There is software out there that will play one file, then another and then randomly one of those two again and you have to listen and say which of the two you hear. I'd bet a good amount of money that no one on Earth could score much higher than a person who had guessed all the answers in advance to day before.

    Get this software
    http://emptymusic.com/software/ABXer.html
    Then put in an ALAC and an AIFF of the same music and do 100 trial tests.

    Then try and compare 256AAC and ALAC. I doubt you can tell the difference if though in this cue there is a difference. Then try 320K and 128K files
     
  7. macrumors 68020

    Prodo123

    Joined:
    Nov 18, 2010
    #7
    ABX test, easy to tell the difference between lossy and lossless. I mentioned this in my other post that I'm sure you read.
    (you seem to have a firm belief that it's not possible when it clearly is)

    Between compressed lossless and uncompressed there is zero difference in sound quality, thus an ABX has a big possibility to give erroneous results as affected by confirmation bias. Since the tester is trying to listen for differences, there will be bias toward "there is a difference between A and B." Furthermore, by a stroke of luck you could choose 9 AIFF in a row and 1 ALAC, creating random statistical bias. Either way it can give the false impression that you can tell apart uncompressed lossless from compressed.

    I mention my decompression speed comparison because it gives concrete evidence that buffer overfill does not occur. It's empirical and simple; there isn't any human factor in how fast a computer can decompress a file, and it's a direct observation of decompression speed vs. playback speed.
     
  8. thread starter Guest

    Joined:
    Apr 17, 2005
    Location:
    Currently in Switzerland
    #8
    I wasn't expecting such emotional answers - I just read the white paper and presented the question here, PARTICULARLY since the Computer Audiophile forum does show diverging opinions from very knowledgeable people.

    Honestly, I did not understand Prodo's example very clearly and would like further clarification as to how he calculated the decompression needs for an AIFF file compared to disk/memory bandwidth...in other words, if decompression MAY somehow be affected by disk speed, CPU or RAM limitations, there MAY be an impact on sound quality...or not?
     
  9. Prodo123, Jul 22, 2013
    Last edited: Jul 22, 2013

    macrumors 68020

    Prodo123

    Joined:
    Nov 18, 2010
    #9
    The difference between compressed lossless and raw lossless is that one obviously requires decompression. Once decompressed, the resulting data is the same, like a ZIP file. That decompressed data is the thing being played, so there is no difference whatsoever between compressed and uncompressed in terms of sound quality.

    Then where does this "Uncompressed has better sound quality" argument come from? It's an old issue in technology when computers were too slow to keep up with the decompression of compressed lossless audio. In a YouTube video, if the video doesn't load faster than the playback speed, then the video will buffer and create interruptions in the video. Likewise, if the decompression of compressed audio can't keep up with the playback speed, the audio will stop until the next segment of audio is decompressed.

    In the old days the computers were too slow to keep up with compressed lossless audio files, resulting in buffer artifacts during playback. Nowadays computers are WAY faster, and decompresses audio faster than playback speed. So yes, decompression artifacts result from a slow CPU if you were wondering. You will only encounter it if you have something like a 1Ghz Pentium 3 from 12 years ago, or have some heavy duty computer rendering that takes up 99.9999% of your CPU resources already. Either way you will never come across decompression artifacts in normal use.

    The example where you convert ALAC to AIFF is just a display of how fast decompression actually occurs. If it takes 2 seconds to decompress a 4 minute song, why would the audio lag? Since lag only occurs if the decompression can't keep up with playback (i.e. it takes longer than 4 minutes to decompress a 4 minute song), one will not come across artifacts.

    I agree with ChrisA in that whatever difference that poster is hearing is NOT real. He's psychologically inclined to "hear" or create differences in the audio files.
     
  10. thread starter Guest

    Joined:
    Apr 17, 2005
    Location:
    Currently in Switzerland
    #10
    Tks; but I still see that players such as Audirvana Plus DO take a considerable amount of time (i.e., not as lightning-fast as you seem to imply) when buffering-playing high bit-rate files such as FLAC or the like - just look at the progress bar of their player when a file is clicked for playback...still fast enough, but not as immediate as one would prefer.

    And I am talking here about my iMac with 16GB RAM, SSD and 3.4 i7; not a Pentium.
     
  11. macrumors 68020

    Prodo123

    Joined:
    Nov 18, 2010
    #11
    That's just bad programming.
    There is no such thing as a "better audio engine." There is one way to decode a file, and that's through a codec. The way that iTunes decodes ALAC is the same way Audirvana decodes ALAC; the same applies to FLAC, etc. That's one of the snake oil marketing that audiophile products use.
    Audirvana is most likely applying some sort of filter before playback. I'd shy away from anything that alters lossless audio, because once a filter is applied it's no longer lossless.
     
  12. macrumors G4

    Joined:
    Jan 5, 2006
    Location:
    Redondo Beach, California
    #12


    First of all the assumption is wrong. It actually takes LESS resources to play the compressed file. There is only half as much data to read off the disk. Reading data is "harder" or slower then decompressing it Reading only half the data is such a huge "win" that it turns his argument upside down.

    But really, in real life there is ZERO difference. None at all.

    You can measure the amount of CPU and RAM it takes to decompress a file, Look at Activity Monitor and you find it says nearly zero.

    You can answer the question of which sounds best by experiment. Get that ABX software and do 100 or so listing tests.
     
  13. macrumors 68020

    Prodo123

    Joined:
    Nov 18, 2010
    #13
    Well that's just erroneous information. Raw PCM does not need a codec thus bypassing the need for information processing. Once the header of the file is read the rest of the data can be fed directly to sound cards. Therefore uncompressed audio is purely limited by disk speed, which is not an issue at all since the entire file can be read in one second on the slowest laptop hard drives.
    Whereas compressed audio needs to read a big portion of data from the disk (albeit half as much), be stored on the RAM, be processed by the codec and be stored in the CPU cache, until finally reaching the sound card.

    Like I said before, ABX is only useful if you know for sure that there is a difference between the two files, as is the case between MP3 and FLAC. It creates false bias and assumption when the two files contain identical data, as is the case between FLAC and WAV.
     
  14. macrumors 65816

    Joined:
    Jul 7, 2008
    #14
    That would confuse anyone. ABX is the worst possible methodology one could choose for audio. Virtually everyone listens to music for satisfaction, so any methodology that doesn't measure satisfaction misses the point completely.
     
  15. macrumors G4

    Joined:
    Jan 5, 2006
    Location:
    Redondo Beach, California
    #15
    If you literally can NOT hear the difference between two files even with 100 attempts spaced as slowly as you like how can one of the files give you more "satisfaction"?

    Or is it that maybe yo simply like the idea of the bits being written to the disk in a certain way, one way is more elegant or artful even if the sound is the same. Kind of like driving over a road you KNOW is build partly from used contraction materials, even if the road is the same it makes you feel good the material was not put in a land fill.

    So maybe file format is like that FLAC is "open" so it makes you feel better?

    Just trying to see why one might care if the sound is bit per bit identical and you know you can't hear any difference.

    ----------

    Got numbers? I'd like to see how bias is calculated. And then if it is a "false bias" I like to know what the number should be

    When the files are identical like FLAC and WAV what I'd expect an exact 50/50 split. The same as if you flipped a coin. (That is 50/50 within the range of experimental error)

    I would not expect a bias of any kind, the ABX test measures you abilty to know which file is which and you should to basically fail the test.
     
  16. macrumors 68020

    Prodo123

    Joined:
    Nov 18, 2010
    #16
    No, ABX tests collect the statistical rate of you choosing the "correct" choice over the "wrong" choice. ABX tests by its inherent nature cannot rid itself of the statistical bias that may occur.
    Here are five different trials between AIFF and ALAC that I had done couple years ago.
    Trial 1: 24 ALAC, 26 AIFF
    Trial 2: 43 ALAC, 7 AIFF
    Trial 3: 34 ALAC, 16 AIFF
    From the raw numbers one might assume that I am able to tell the difference between ALAC and AIFF, which is completely wrong.
    Also, flipping a coin also has the possibility of statistical bias; if you're lucky you might get 70 tails and 30 heads, leading to the false assumption that the tail side weighs less than the head side.
     
  17. macrumors 6502a

    Joined:
    Oct 4, 2008
    #17
    As far as your audio gear is concerned, there is no difference between ALAC and AIFF. Both produce the exact same signal at the socket on your computer.

    As far as your computer is concerned, ALAC requires more resources to decompress and play then AIFF does, at the cost of disk space (ALAC files are obviously smaller then AIFF files). Computers (and handhelds) are so friggin' fast today that the ALAC bit doesn't matter, which leaves AIFF with the only disadvantage- taking up a lot of disk space for absolutely no advantage whatsoever.

    If you really want to compare them, import an ALAC and AIFF file into a DAW application like Logic 9, Logic X, Cubase, Protools, whatever. Invert the phase on one channel, crank your speakers and whack play. You will hear nothing, signifying that the two files null out completely which means they are identical. If you have a scope plugin, you can always verify this with a spectrogram or similar.

    ALAC is lossless, end of thread. If you think AIFF is somehow better then ALAC, then I have another wonderful device I'd like to sell you.

    -SC
     
  18. fa8362, Jul 24, 2013
    Last edited: Jul 24, 2013

    macrumors 65816

    Joined:
    Jul 7, 2008
    #18


    I don't care about bits. I'm simply telling you that ABX testing is an inappropriate methodology for listening assessment. It doesn't even answer the question that is being asked. The question is which variant sounds better, not which variant is which.

    Do you think that candy manufacturers determine which candy formulation variant tastes better by having respondents switch back and forth between the variants??? And then question them on which variant is which??? Answer: They DON'T. Why? Because they know better, and which is which is irrelevant to what they're trying to determine.
     
  19. macrumors 65816

    Joined:
    Jul 7, 2008
    #19
    My advice to you is to stop reading about what others think, and do your own listening and decide for yourself.
     
  20. macrumors 603

    Joined:
    Jun 30, 2007
    Location:
    Atlanta
    #20
    While I don't disagree with your assessment I will take on your analogy. Candy (or any food) taste is 100% subjective. Asking how an ALAC or FLAC compare to the "gold standard" original master is 100% objective.

    In a nutshell ALAC and FLAC are 100% bit for bit identical after decoding to the original master used. No ABX or even more accurate DBT are needed because it IS a fact the files are identical.

    No one would ever challenge the accuracy of a book or spreadsheet that had been Zipped compressed and the same should hold true for an ALAC or FLAC file.
     
  21. thread starter Guest

    Joined:
    Apr 17, 2005
    Location:
    Currently in Switzerland
    #21
    Tks, but my question is a little bit different. I DO know both are lossless; however, one is compressed, the other is not.

    So in the end I simply want to know if there IS a scientific possibility for AIFF to sound better DURING PLAYBACK since it does NOT tax CPU or disk with decompression routines, particularly in scenarios where you are doing many other things at the same time.

    I repeat: Audirvana Plus's playback bar shows that it takes a certain time to read a whole file in ALAC or FLAC - and other people have reported in specialized forums that AIFF does sound better...so what you are all saying is that this is just placebo or snake oil perception?

    Tks again for your feedback.
     
  22. fa8362, Jul 24, 2013
    Last edited: Jul 24, 2013

    macrumors 65816

    Joined:
    Jul 7, 2008
    #22
    The perception of what one hears is NOT objective. It is subjective. The brain is involved, and can't be eliminated from the equation. If people claim to hear something, that's their reality. No amount of arguing, foot stomping, or nonsensical ABX testing is going to change that.
     
  23. macrumors 603

    Joined:
    Jun 30, 2007
    Location:
    Atlanta
    #23
    True but...the FACTS are 10010011001001001010 in LPCM is 100% the same as 10010011001001001010 in decompressed ALAC. Someone may THINK they hear a difference (or think the word is flat) but that is a FALSE belief. One's belief may be one's truth but that doesn't make it a fact.

    You can have a TRUE subjective opinion on lossy codecs (like AASC or MP3) compared to the LPCM Master. However any subjective opinion anyone thinks they hear on differences in LPCM vs ALAC (decoded) is factually wrong.;)

    Also CPU cycles for decompression have NO bearing on SQ. The LPCM stream (after decoding) is tied to a master clock and is feed out at the same data rate (and 100% bit for bit) as a straight LPCM stream. If for some reason the CPU could not keep up you would have dropouts and not true quality reduction.
     
  24. macrumors 6502a

    Joined:
    Oct 4, 2008
    #24
    No, there is no scientific possibility.

    The CPU is too fast for that to be an issue. You're loading up data into a RAM buffer (and optionally decompressing it) in the blink of an eye, while a separate thread is busy actually playing that data out whatever audio port you've chosen.

    Since the actual loading of the data occurs faster then that audio can be played back, it is impossible for AIFF to "sound better" then ALAC.

    Again, you can test this yourself by performing a null test. Or you can hop on an actual audio forum like Gearslutz.com and ask those guys- you'll get a hell of a lot more technical discussion about this issue then you will here, but I imagine the answer will be the same- as far as your outboard gear is concerned, ALAC and AIFF are empirically identical.

    -SC
     
  25. steve-p, Jul 25, 2013
    Last edited: Jul 25, 2013

    macrumors 68000

    steve-p

    Joined:
    Oct 14, 2008
    Location:
    Newbury, UK
    #25
    None whatsoever. The puny processor in ancient iPods copes fine with decompressing ALAC files in real time, so a modern desktop or laptop CPU could probably process hundreds simultaneously without issue if the disk could keep up, and 16/44.1 audio is only 176 KB per second even when uncompressed. ALAC is typically ~40% less.
     

Share This Page