|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
75.214.6.212
In Reply to: apple lossless is best posted by Magpie on December 4, 2006 at 19:57:50:
I find the Apple Lossless to attenuate inflections and vibratos by the performers. It was kind of spooky when I tried it.
Follow Ups:
Todd,Can you please describe the hardware/software setup that you used to conduct the comparison of Apple Lossless file playback performance versus CD?
Apple lossless is totally lossless, the decoded file matches bit for bit the original CD. With a hex editor you can do this test that proves it.
1. Take any 16-bit .wav file and import it into iTunes (you can rip it from a CD with iTunes)
2. burn it to a CD
3. re-import it from the CD using apple lossless encoding
4. reconvert it to a .wav file
5. use a good hex editor to calculate the CRC (checksum) of the two .wav files, they will match exactlyA CRC is a cyclic redundancy check, it uses nonlinear division to divide the file by a (usually 32 bit) number and returns the remainder, the likelihood of the files CRCs matching if the files are not identical is 1 chance in 2^32, or 1 chance in ~4 billion so you can have reasonable confidence that audio files with identical CRCs are truly identical. Interestingly there are very very few CD ripping/burning programs that do not change the audio, iTunes is one of the few that is totally bit perfect, as long as you use lossless files.
Yes, I know that files ripped using Apple lossless result in output that's bit perfect compared to the original source file. That's why I'm rather incredulous that an Apple lossless file can be audibly different from the original. It would be most illuminating if Todd could share some details regarding the methodology he used for his listening test, but while I'm curious about what his answer could be, I'm not going to indulge in speculation.
> > an Apple lossless file can be audibly different from the original.It can't. It can only be alleged due to open-loop subjective listening tests.
HowdyDoesn't that depend on whether the decompressing causes untoward effects in the downstream components, say noise on the power supply, buffers late for the DMA, etc.? After all we still hear people giving advice to not have too many other apps, drivers, etc. running when you are playing audio. Granted in an ideal world with ideal implememtations these things wouldn't be a problem, but practically speaking they often are and reputedly can cause audible effects. (I say reputedly since I don't listen to music from my PC very much and have no personal experience with Apple lossless.)
"Doesn't that depend on whether the decompressing causes untoward effects in the downstream components, say noise on the power supply, buffers late for the DMA, etc.?"Given that even a low-powered device like an iPod can decode an Apple Lossless encoded file, the overhead associated with this task is very low indeed. For current PCs and Macs, with their brute power, it would be trivial.
"After all we still hear people giving advice to not have too many other apps, drivers, etc. running when you are playing audio."Yes, but we need to be specific here about what we're dealing with. The issue here is that high-overhead tasks (such as anti-virus scanning and DVD video playback or encoding) will consume significant CPU and memory resources, thus making the performance of other background tasks suffer as your computer literally bogs down from the load. The key thing to note here is that Apple lossless decoding by itself is not a highly computationally intensive task (far from it) but if you are multi-tasking many other apps or running some other job(s) that heavily taxes your computer's resources, then your computer's overall performance will be negatively impacted. The fact that your audio playback performance can suffer as a result is one of the symptoms, but not the cause.
I think that if you wanted to test whether lossless compression (and lossless compression alone) causes audible differences in playback compared to the original source file, one way would be to run a double-blind test in which both files are streamed from the hard disk of the same computer (acting as a transport) to feed a DAC. Any way you do it, you would have to play back both files in the same system using the same transport and media, otherwise you have too many variables that can influence the result.
HowdyI was just pointing out the fallacy of Greg's post:
"> > an Apple lossless file can be audibly different from the original.
It can't. It can only be alleged due to open-loop subjective listening tests."
In any case if you can hear a problem with differing CPU loads then there is a problem with your setup and the specifics of the amount of load don't matter that much. Unfortunately experience tells us that CPU load is often audible. There are reports here of people that indeed did A/B tests with lossless compressed vs. uncompressed with all other variables the same. As always the people who don't believe there is a difference don't accept that the experiments are controlled enough and the ones that do believe there is a difference don't care about anyone else's experiments :) Anyway I did my experiments before the IPod, I don't have an IPod and I don't really care :)
By the way, I should mention that my interest in Apple Lossless stems from my own attempt to build a computer-based music server using a Mac and iTunes as the engine. My choice of hardware and software is based simply on personal preference -- I just prefer to deal with Mac OS rather than Windows. On top of that, I've chosen to encode my files in Apple Lossless in order to achieve an optimum balance between sound quality and storage efficiency. In my own [admittedly crude] experiments, I've been unable to detect any audible difference between streaming music in Apple Lossess format compared to native WAV format when both were played directly off my hard drive and passed to my outboard DAC. The conclusion I've come to is that lossless compression results in output that is both quantatively and qualitatively equivalent to the original. I personally find it very gratifying when the brain and the ears are in agreement, but that's just my $0.02. ;)
HowdyFrom what I've seen OSX has a pretty nice audio stack. It was a lot harder when we wrote a hard disk recorder for Win 3.0 :) I think a lot of people don't really know what it takes to do good audio in a PC and overlook a lot of the details that matter. When I hear a system that has artifacts of any sort under any CPU load I know that the designers are in over their heads...
Hi Ted,"In any case if you can hear a problem with differing CPU loads then there is a problem with your setup and the specifics of the amount of load don't matter that much. Unfortunately experience tells us that CPU load is often audible."
Yes, and when it happens it manifests itself as clearly audible dropouts and "stuttering". However, it should be mentioned that this can happen whether you're playing back compressed audio, uncompressed audio, or video for that matter. Therefore it appears to me that we're talking about two separate topics.
a) The audibility of lossless audio file compression.
b) The merits of using a computer as a digital audio transport.
Ultimately I do agree with you regarding the issue of CPU load on audio playback performance, but I think that this speaks more towards topic "b" than topic "a". For this reason, I do not see any fallacy in Greg's post. It's obvious that a computer can be a perfectly capable audio transport as long as you don't overburden it with concurrent tasks. Once this simple rule is heeded, issue "b" shouldn't be a worry. However, the validity of issue "b" does not presuppose the existence of issue "a".
"There are reports here of people that indeed did A/B tests with lossless compressed vs. uncompressed with all other variables the same. As always the people who don't believe there is a difference don't accept that the experiments are controlled enough and the ones that do believe there is a difference don't care about anyone else's experiments"
Your last point hits home. We could continue debating the point of whether lossless compression has audible artifacts, but I'm really not inclined to do that. In the end, I've said my part and I respect the rights of others to state their opposing opinions. It's only when matters are presented as fact that I think we should ask for a higher burden of proof to be presented.
This post is made possible by the generous support of people like you and our sponsors: