--- "Paino, Christopher L YN1(AW) (CPF N0084)" <PainoCL@cpf.navy.mil> wrote:> If I am understanding FLAC correctly, the internal MD5 sum is (or can > be > )different than the fingerprint file sum.Yes.> The internal sum can be > different > depending on which compression settings are used, while the audio > remains > unchanged.It's the other way around. Let me sum up my limited knowledge about how stuff is seeded: Shorten: UserA starts with WAV file UserA compresses with shorten, default arguments UserA creates fingerprint by running md5sum on the SHN file (If the user recompresses with different options or a different version of shorten, the resulting shorten file will be different causing the md5sum to be different.) UserA makes a text file with the MD5 sum and SHN filename UserB downloads text file and SHN file, runs md5sum on it, and compares it to the text file. FLAC: User starts with WAV file User compresses with FLAC, any arguments, doesn't matter FLAC encoder stores MD5 of uncompressed audio internally (If the user decompresses the FLAC to WAV and reencodes, even with different versions of flac or different options, the MD5 sum will be the same). User makes a text file exporting the internal MD5 sum from the FLAC file, and the FLAC file name. UserB downloads the text file and FLAC file, and test it using flac -t, then exports the MD5 sum from the FLAC file, and compares it to the text file. Assuming this is all correct, the extra step for FLAC is exporting the MD5 sum to compare to the text file, because flac -t doesn't do it. It's true this is an extra step than with shorten, but the tradeoff is that with FLAC the fingerprint survives reencoding, and it verifies that the encoding was done properly in the first place. In any case, if you run flac -t and it passes, the only extra information you get from comparing the MD5 sum to the text file is to know if the original seeder put the right text file together with the right FLAC file. But if you don't need the contents of the text file to know if the file was d/l'ed OK, why do you need the text file at all? Josh __________________________________________________ Do you Yahoo!? New DSL Internet Access from SBC & Yahoo! http://sbc.yahoo.com
--- Josh Coalson <xflac@yahoo.com> wrote:> --- "Paino, Christopher L YN1(AW) (CPF N0084)" <PainoCL@cpf.navy.mil> > wrote: > > If I am understanding FLAC correctly, the internal MD5 sum is (or > can > > be > > )different than the fingerprint file sum. > > Yes.Let me clarify that... the internal MD5 sum is different than what you get from running md5sum on the FLAC file. I believe which type fingerprint actually goes in the fingerprint file is still under discussion within etree. Josh __________________________________________________ Do you Yahoo!? New DSL Internet Access from SBC & Yahoo! http://sbc.yahoo.com
Our Leader asked, | In any case, if you run flac -t and it passes, the only extra | information you get from comparing the MD5 sum to the text | file is to know if the original seeder put the right text file | together with the right FLAC file. But if you don't need the | contents of the text file to know if the file was d/l'ed OK, | why do you need the text file at all? [Anywhere in here I say "fingerprint," I mean the internal one.] As it has been explained to me, the original seeder keeps a file of fingerprints available by FTP or HTTP, so that anyone down the chain can compare the fingerprints of his/her FLACs to those created by the seeder. If someone along the way has reencoded the same music with different FLAC options, the fingerprint won't change, so there's nothing to be concerned about. However, if the recipient's FLAC files pass flac -t but the fingerprints don't match those on the seeder's site, somebody along the way has changed the underlying WAV data. That's a two-edged sword: it stymies a malefactor from altering the WAVs out of mischief, but it also prevents a music lover from adjusting track marks, declicking, or degapping.
--- "David W. Tamkin" <dattier@panix.com> wrote:> Our Leader asked, > > | In any case, if you run flac -t and it passes, the only extra > | information you get from comparing the MD5 sum to the text > | file is to know if the original seeder put the right text file > | together with the right FLAC file. But if you don't need the > | contents of the text file to know if the file was d/l'ed OK, > | why do you need the text file at all? > > [Anywhere in here I say "fingerprint," I mean the internal one.] > > As it has been explained to me, the original seeder keeps a file of > fingerprints available by FTP or HTTP, so that anyone down the chain > can > compare the fingerprints of his/her FLACs to those created by the > seeder. If > someone along the way has reencoded the same music with different > FLAC > options, the fingerprint won't change, so there's nothing to be > concerned > about. However, if the recipient's FLAC files pass flac -t but the > fingerprints don't match those on the seeder's site, somebody along > the way > has changed the underlying WAV data. That's a two-edged sword: it > stymies a > malefactor from altering the WAVs out of mischief, but it also > prevents a > music lover from adjusting track marks, declicking, or degapping.I thought once you change the audio it requires a new seed. Josh __________________________________________________ Do you Yahoo!? New DSL Internet Access from SBC & Yahoo! http://sbc.yahoo.com