First of all, I am not sure if I should call this a test. I dont know if it is rigorous enough and I dont have enough theoretical and practical knowledge so that my conclusions worth anything. I am at best novice in this field so errors are quite probable Anyway, I will provide you the raw data and observations and you can repeat the tests and draw your own conclusions. After a long discussion about what source is better: the CD or a file in the computer, I started comparing a few different ripping programs. The files had been saved as wav, so no compression was implied. For ripping and saving the files, I used: - the most common Windows Media Player version 11.0.5721.5260 - an old and quasi professional Sony SoundForge version 8.0d (Build 128) and - the much praised Exact Audio Copy V0.99 prebeta 5 from 4. May 2009 For all the programs, I used the best quality settings. At WMP there were no much settings. I chose quality instead of compression. At SoundForge I used 1x speed and for EAC I used both the secure mode and the paranoid mode. Even the program doesnt recommend the paranoid mode. In this mode, the program does even override the buffer of the reading device. While doing this the ripping could get a lot longer than usual. In fact, in paranoid mode ripping took several hours!!! This mode is recommended for damaged CDs. But I think it will take less time to dress yourself, get in the car, drive around town and find a place to buy another CD, get back home and do the job. At some point I played solitaire and watch TV from 8 36 in the evening until 12 23. I was ripping Led Zeppelins Good times bad times. It is a 2 min 46 seconds song. And the CD was NOT scratched. But the program doesnt care. It was doing its endless checks as it was supposed to do. Paranoid mode is a very well chosen name! To cut the long story short, in the end I was having several copies (four) of the same track. One WMP generated wav, one SSF and two EAC. One was generated in secure mode and the last in paranoid mode. I must admit that I didnt do a listening test on a decent audio rig. Therefore, the sound was absolutely the same to me no matter what program generated the wav and no matter the program was playing it. However, and you know that better than me, listening is a lot more complex process. I wanted only to compare the digital copies. To compare them, but how to do that? The simplest way was to compare the length of the file. Obviously, it was not the same. Further more I used 7ZIP (4.45 beta) to calculate the check sum. Again, the numbers were different. However, the EAC paranoid and secure wavs were the same. But that was to be expected. As long as both modes (secure and paranoid) were supposed to extract the perfect copy, it was logical to obtain the same - I mean identical file. As a next step, I used the EACs feature of comparing the files. Again, as expected, its own secure and paranoid files were found identical. But when I compared any of those files with a WMP or SSF generated wavs, differences begin to occur. The WMP generated wav had 667 repeated samples at position 0:00:00.388 over the EAC generated wavs The SSF copy had only 657 repeated samples, but at the same position 0:00:00.388. Now comparing the SoundForge generated wav with the Windows Media Player generated wav the answer was: WMP wav has 10 repeated samples aver the SSF wav at position 0:00:00.403. At this point, the EAC analysis makes sense. If the WMP copy has 667 repeated samples compared to EAC copy and SSF copy has only 657, it is logical that WMP will have 10 repeated samples over the SSF copy. We may explain the change in position by the fact that the repeated samples the WNP copy has over the SSF copy are located at a later position. Somewhere after position 388. (Maybe between position 388 and position 388 plus 667.) This is true only if any sample corresponds to a single position. If a position contains more than one sample or a sample is located over more than one position, the change in position of the error still makes sense, but we cant control its location. Maybe someone knowing more about samples and positions can understand better and certifies if this explanation is OK or not. The first question marks were raised by the fact that at those comparisons EAC also stipulates file 0:00:00.000 longer. I see no meaning for that. Longer with zero! It was not a random defect. This occurred at every comparison of the same tracks. EAC copy was 0 longer than SSF copy. EACs copy was NOT longer than WMP's copy and, logically, WMP's also appears longer than SSF generated wav. Zero longer that is! Ok! This is just something I dont understand. But, then, the real question marks were raised when doing the same tests to ABBAs Does your mother know?. Instead of SSF and WMP copies having repeated samples, ABBAs EAC copies were missing 667 samples! The repeated samples were at position 0:00:00.000. Again 667?! Again 0?! If the number of samples would have been 666, I would have started suspecting thighs here! Anyway, past the number coincidence, there is nothing to talk about. Those copies were missing some samples while the others were repeating some samples. The news is that EAC found no difference between the WMP generated wav and the SSF one. So, what we have until now? Better said: what we dont have! We dont have three same wav files! At Led Zeppelin, all three rips generated by the three different programs were different! At ABBA, according to EAC analysis, the wav files generated by the Windows Media Player and SoundForge were the same, but different then the file generated by the Exact Audio Copy. At this point, I turned to WinMerge (version 18.104.22.168) But the WinMerge doesnt compare wavs. WinMerge compares files. This includes the name of the file and other information. These bits are not directly related to the music included in the file. And here is the biggest surprise! Except some lines at the beginning and at the end, WinMerge found the files identical. In fact, at some files, only lines at the beginning were different. The rest of the file was identical to the last bit! What are my conclusions? Remember that the differences found by EAC were at positions 0 or "0 something"? ( 0:00:00.388) So this is less than 4 tenths from the beginning of the file. The rest is the same. In my opinion, ripped trough Exact Audio Copy, Sony SoundForge or Windows Media Player, the bits of the music will be the same. And the ripping is consistent as I obtained the same results ripping several times in any configuration. So the ripping program seams not to be important as long as we set it for quality. If anybody obtains different results and/or observes flaws in my deductions, I am all ears. As I said, I am at best novice in audio so errors are quite probable. And I am also willing to learn more. And here is the final and biggest surprise at all: The same song from different albums generates completely and entirely different file! Good times, bad times from the very best of Led Zeppelin 2 CD box 0 75678 36195 1, CD no 7567-83619-2 vol1 Atlantic... produced a completely different file than... ...the same song from Led Zeppelin vol IV CD no 7567-82632-2, also from Atlantic. Both CDs where digitally remastered from the same original tapes, bla-bla, bla-bla... But this is another story I presume!