|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
75.212.253.222
In Reply to: digital compression music test posted by Gulliver on April 11, 2007 at 12:10:40:
I think the biggest reason people "fail" CD-vs-MP3 listening tests is that the conduit for running such tests often does not maintain the signal integrity, which mostly impacts the sound of uncompressed signals. Another problem is that a lot of purportedly "uncompressed" data that is not on original media has been compromised.The only way to run a test like this properly is to do an ABX/DBT with the uncompromised uncompressed signal vs. the encoded/decoded compressed signal, burned onto CD-R, on a playback system that does not compromise the integrity of the Redbook signal. (About 90 percent of CD playback systems *do* compromise the Redbook signal, IMHO.)
I plan on doing this with non-compressed vs. FLAC or Monkeys Audio. Burning an "ABX CD". Using dBPoweramp as the ripper, utilizing different encoding/decoding within the ripper. (The non-compressed tracks will be done with no compression in data transfer.) I think the sound is degraded not only with classic MP3 compression, but so-called "lossless" compression as well. And I think it can be proven with a DBT, under tightly-controlled conditions.
Once I finish testing, I will make this "ABX CD" available to fellow inmates, simply to allow independent verification of the data. The information given should result in verification via comparison utilities like EAC.
Follow Ups:
z
There are more and more products out there where "lossless" compression is part of the package, where a lot of people presume comparable or improved performance. While at the same time, another group of people claiming that we cannot distinguish the difference between Redbook and MP3, with nobody realizing that the "Redbook" may in actuality be degraded ("lossless"), and a lot of false belief that maybe MP3 is all we need. So what we end up getting is even more compromise. Not just in poorer sounding products, but lower quality recordings.There is so much confusion over what's a good or degraded signal, we end up buying what's purportedly "new technology" that is in reality new compromises in performance.
...that you cannot distinguish between a lossless version and the original. I have done quite a bit of listening to FLACs vs. WAVs and I am completely unable to distinguish a difference. This allows me to store almost twice as much music on my server and makes me quite happy. Of course, YMMV...
I think the difference is audible to where it would pass a DBT. I plan on doing this the weekend following this weekend.I burned a CD using ripped with EAC using FLAC 1.1.4, because it was supposedly an optimized setting according to some of the pro sites. (I optimized all the settings, including the offset for the Plextor 708 drive.) I wanted a travel recording of a CD I had. I briefly played it on the main system, and wondered what went wrong. I then burned another CD ripped with EAC using the ancient Microsoft PCM conversion. I could not believe how much better the second disc sounded. Same ripping software. Same burn software. Same hardware. Identical media. Same settings, except for the encoding (FLAC vs. direct PCM) from the ripped CD. (I later tried dBPowerAmp, and prefer it over EAC. I can choose PCM or FLAC with that ripping software as well.)
I figured it shouldn't be too difficult to burn a CD with the tracks side by side. So it's even the same disc. The only thing I'd have to do is maybe shorten the tracks to make such comparison practical. It shouldn't be too difficult to do. Just the right tools need to be utilized.
...and boy were we surprised!
That when you listen to a FLAC file natively, you are listening to the re-construction of the WAV file on-the-fly. Therefore, you are at the mercy of the software being used to play back the FLAC file. The real benefit of FLAC is to allow for the storage of more files on the hard disk. If you want to listen to the WAV file in all its original glory, you'd need to convert the FLAC to WAV.Clark, are you and your friends willing to submit your FLAC hearing acumen to a DBT?
I didn't think so....
"Clark, are you and your friends willing to submit your FLAC hearing acumen to a DBT?"I only speak for myself, but if you read the thread, this is *exactly* what I plan on doing. (I guess "ABX" is a subset of "DBT", but the intention is the same.)
I think there is an audible difference. The test will be in the form of an "ABX" test on CD, where I listen to the uncompressed and "compressed/restored" tracks as "A" and "B", and will have to determine which version "X" is. On an 80-minute CD-R, I should be able to get 25 or more trials burned onto one disc. (I might do two CD-Rs.) I will log whether "A" or "B" is "X", and then compare what I hear with what "X" actually was for those trials after listening is finished.
The tracks will span classical, jazz, rock, and other genres. But no pitch-corrected vocals (that eliminates a *lot* of artists), no tracks devoid of acoustic instruments, no phase-inverted tracks (I might flip the polarity if need be), no "mixed polarity" tracks, and no tracks whose original resolution is lower than CD-quality. For the sake of playback compatibility, no HDCD or pre-emphasized tracks will be used. The intention is to use the best fidelity possible, where any compromise in resolution could also compromise the listening experience.
This will be done on a PC with 2.8 GHz Pentium 4, Windows XP Service Pack 2, FLAC 1.1.4 compression algorithm, MS PCM for the non-compressed conversion, dBpowerAmp for the ripping software, Plextor 708 CD/DVD RW drive with the latest firmware, Prassi ONES burn software, and Mitsui Gold Audio (1x-12x) CD-Rs. Burn speed will be at 4x. Some of the software or media might change when the test is actually conducted.
I expect to conduct the test within a couple months. Several copies of the "ABX" CDs will be available free of charge to the those quickest to respond. I might request others to conduct the test. (I will not post results of anybody's individual findings without their permission.) My results, good, bad, or ugly, will be posted here.
x
nt
I notice it with Apple Lossless, FLAC, Monkeys Audio, LAME, and Ogg Vorbis. (I will also try WavPac.) The flavors are different, but all lose that absolute resolution that is so difficult to attain to begin with.If I was forced to choose a lossless algorithm, I think FLAC does the least damage. Monkeys is almost comparable. At least I don't hear some wacky artifacts like I do with Apple Lossless or LAME.
I can go back and look at my notes, if you'd like.
This post is made possible by the generous support of people like you and our sponsors: