I'm not a 'self-proclaimed' audiophile I'm a musician and a producer and readily agree that 128K AAC is a clean bright and polished sound, equivelant in my opinion to 256K MP3. I will admit it's very difficult if not nearly impossible to discern artifaction but when you listen to a 128K AAC and then the exact same file at Apple Lossless ( or even 256K AAC) you will notice a difference in presence. All of the tangible differences are incredibly subtle but you tend to notice it in reverb, depth of soundstage and the 'tonal' quality of the treble and just the general cohesiveness of the mix...
I appreciate the feedback but anecdotal evidence (regardless of who its from) is next to worthless when speaking of differences that should be objectively discernable not just by anyone but by any system designed to quantify such discrepancies.
I too am a musician and a producer. I too use sophisticated equipment. I too master recordings at 24-bit depth. I too hear things other people do not hear (the 60Hz hum in a cathode ray tube while barely discernible to most is glaringly present to the point of annoying me whenever I walk into a room where someone's left a display on with the sound off).
And yet I will bet that you will not be able to correctly identify the 128kbps AAC track in a double-blind test more than 75% of the time. Say whatever else you want, but if you cannot readily identify it more than 75% of the time, then the consistency of your perception, not the fidelity of AAC, needs to be called into question.
A double-blind test requires that neither you nor the person administering the test know which samples are AAC and which samples are PCM. The samples must be repeated enough times to ensure statistical relevancy, and must also be weighed against the results of your selections in a 2nd trial where you are told what the formats are, and repeated again in a trial where you are given false information (i.e. told it's AAC when it's really PCM and vice-versa).
I will also bet that you will frequently erroneously identify the PCM files as having poorer fidelity when you are told falsely they are AAC files.
I listen to my music through either a very expensive set of studio monitors which are designed to make music sound BAD ( or at the very least reveal flaws and imperfections that consumer friendly speakers do not) or a pair of expensive studio headphones.
There are many expensive systems that are mediocre. Price is not always an indicator of great fidelity.
To wit... $3500 Wadia CD players are just repackaged Pioneer transports with Burr-Brown DACs. Granted, the Burr-Brown DAC can make a significant difference but I find it funny that there's a bunch of people out there buying $3500 CD players who would probably thumb their noses at Pioneers.
I will 100% admit you can't hear the difference on iMac speakers, cheap stereo or even ipod headphones but I assure you despite what the scientific graphs and boffins declare the sound is not as pure.
Your assurances mean nothing if your perception is flawed.
I still don't see any evidence of you understanding the fundamental principles of digital audio encoding. Whenever I hear the "AAC is crap" arguments I never, ever, hear from the opponents a dissection of the encoding schema for AAC that might actually result in substantially higher (read: measurable) degrees of interpolation error, quantization noise, aliased/foldover frequencies, elevation of the noise floor, frequency response roll-off, etc.
What it seems to amount to is this bogus perception that the same sound cannot be constructed from fewer data without understanding how such a thing is mathematically possible. As a simple example, consider the difference between Linear PCM and Adaptive Delta PCM (ADPCM). ADPCM is a lossless coding schema. It achieves the same result as PCM but with fewer bits. How? ADPCM only records the difference in amplitude, rather than the absolute value, from one quantization interval to the next. This results in a substantial decrease in the potential bits of data required to reconstruct the exact same analog waveform.
AAC goes several steps further by factoring human perception into account. Like AC-3, there are parameters in AAC that achieve efficiency at tremendously low bitrates because there's a lot of information typically encoded in a PCM stream that will never affect how you perceive the intended analog fundamental.
Show me, for example, how an AAC algorithm fails to reproduce the same amplitude value at a given quantization interval and WHY and then we might actually enter into a discussion of the logarithmic scale and whether or not the difference in amplitude value is large enough to be discernible by the human ear. It very well may be, but all the anecdotes in the world don't begin to take us into a real academic discussion of digital reproduction systems.
Have you read Pohlmann, by any chance?
Regardless of any of that - the bottom line is if the customer thinks there is too much lost they will buy a CD - and that's bad news for Apple!. That's where I am now and it annoys me becuase I would love to buy a ton of stuff from iTunes - I love it and I like it's distribution model but the long term compromise is too great. DRM doesn't bother me one bit!
I think Apple need to tackle this problem in a different way ;
ie You download Lossless or 256AAC but you can if you want SELECT by preference the a bit rate of the tracks that get put onto your ipod. This will allow them to retain the '1000 songs in your pocket' concept but allows the customer to hold a pristine archive copy on their mac/iTunes.
Actually I would even pay a 10% premium for lossless if it helps Apple cover bandwidth bills!, so that people who are happy with 128K AAC can still d/l at standard rate and that I would download at lossless and pay another £0.79 per album say...
Again, given that I will contest the notion that 128kbps AAC is fundamentally discernible from 16-bit PCM, my personal answer to all this has been to use 128kbps AAC for all my casual listening. For critical listening, I don't even use Apple Lossless or 16-bit PCM. I use 24-bit linear PCM which produces substantial gains even the average listener will usually notice. There again is a tremendous difference between the dynamic range of 16-bit and 24-bit PCM . Frequency response should never be a problem as long as the proper considerations of A-weighted sound are taken into account. That is to say that any digital encoding schema designed properly would low-pass filter any frequencies above the Nyquist limit so as to eliminate alias/foldover. This ensures that the frequency response of the reconstructed analog wave is identical to the original.
Things like "tonality" and "presence" are also largely a function of the dynamic range of an encoding schema. Certain types of music that push the boundaries of 16-bit's dynamic range are much better suited at 24-bit which is why I'm a huge proponent of media that support 24-bit. The problem in the hardware world is that manufacturers like Sony create proprietary encoding like DSD (1-bit, 2.7MHz) for proprietary media (SACD) and then refuse to make their players compatible with other optical media (24-bit DVD Audio). This isn't as big a problem in the computer world and even iTunes supports 24-bit PCM.
So if you're going to complain about AAC over "tonality" and "presence" then throw away all your CD's because they're essentially just as worthless.
