Par2Par: p2p PAR2 sharing for CD rip repair

[EliC]

I mentioned this idea before and would like to bring it up again. With the goal of making the next iteration of dMC the ultimate and most popular CD Ripper I would like to re-suggest a feature that no other ripper has:

Data Repair/Recovery via PAR2 files

No matter how good a ripper is it cannot recover lost data (ie label side scratches...). The idea would be that users would alot a certain amount of disc space for the storage and sharing of PAR2 files created from their rips. The PAR2 files contain no music data and without the wav file could not be used to re-create the music. When a user attempts to rip a track with unrecoverable errors as much of the wav is ripped as possible. Then the system would use a unique disc identifier to search a p2p network (or central server is someone wanted to host it) for the par2 file to repair the rip. The CRC of the repaired file would then be compared with the AR db for a final check.

The system would require an original pressed cd with the same number of tracks with matching lengths per track, and matching AR checksums for accurately ripped tracks. Because the par2 files contain no music data (esentially its just metadata) and the original disc/ripped wav would be required, there should be no legal issues.

[LtData]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

Unfortunately, repairing the files involves rebuilding data you do not have available, therefore it could be seen that you are "downloading" data to complete your songs and therefore violating copyright law. With the RIAA being so... yea... these days, it is a big risk.

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

that would be the case if you just downloaded the missing part, but I would think that since the par2 file requires the rest of the file and is a mathmatic algorithm to rebuild the data based on the data you have that there would not be a copyright violation. Clearly a lawyer would need to address this and I am no lawyer.

[LtData]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

My point is the file that is downloaded doesn't contain any real data itself, correct. However, it is used to create the rest of the missing file. It could possibly be seen that the file you downloaded "contains" copyrighted data and is therefore illegal.

However, as you said, a lawyer would need to address this and I am also no lawyer.

Also, I do think this is a good idea for when you just can't get a CD to rip properly.

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

you could argue that you could potentially do the same thing with a hash. Say you know that only one bit is wrong and you know what the correct crc checksum is, you could change the bit and keep re-calculating the crcs until they match, then you must have the correct data. Its all very grey when it comes to the law, but I think clearly its within any moral boundries as you clearly must have 99+% of the original data for the par2 to do any good.

[LtData]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

I agree with you totally on the moral side and would LOVE to have this. I'm just playing devil's advocate, that's all.

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

I understand. I was just pointing out how if this argument was carried to the extreme even hashes could be illegal. Maybe the best way to do it would simply be having some sort of plugin architecture and someone could provide the plugin to the ripper.

This could be made even more legit by storing the par2 files only with the name and track # based on the disc ID key and NOT providing any mechanism to search for the par files except having the original disc and using that to calculate the disc ID key.

[bhoar]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

It's a neat idea, but there are lots of missing questions...

What chunk size would you propose? Would track boundaries be involved or no? What percentage of missing data would the service need to store? And who would provide it?

The storage aspect is troubling me...unless...hmm.

Ah. I could see it working better if it were a P2P network that autogenerated the necessary parity based on the currently stored/verified disc images of participants, but also prevented users from downloading more than some %age of parity data. Whether that would increase or decrease the legal risks, I am not sure.

-brendan

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

It is a very rough, idea/proposal.

"Damaged files can be repaired. A single byte error in a 10MB file might only requires the use of recovery data from a PAR file that is only 100KB in size."

I think a rough idea of the average wav size from a large db sample, and an average number or errors or size of errors would help to decide how big of a correction file to create.

There must be a balance between the size of the correction file and the amount of data that could be corrected. If the above numbers are linear and the average wav is 50Mb, then to correct 1 byte you would need 500KB. Is 1Mb enough to correct the average number of errors? Anything greater then 5Mb/song is probably asking users to store to much.

As I suggested above, the best thing to do would be a p2p system. You would need to get the right pressing of the disc. To do this you would search on a number of disc features: number and length of tracks, CRC match of other tracks that had an accurate rip, track gain (from replay gain calculations) could all be used to identify the correct disc/pressing on the p2p network


[COLOR=Red]Spoon, could you add a reporting of data from the new CD ripper to start building a db of average wav size and average number of errors in songs with errors?[/COLOR]

[Spoon]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

Reporting to where? a log file?

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

clearly it would need to report somewhere centrally. This data could be collected when AR results are reported.

[Spoon]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

It would need a fundamental change to accuraterip (it has no interaction with CD drive so does not know about c2 errors, etc).

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

it would not have to be reported w/ AR, but certainly r12 will have this data and could easily auto-upload the info.

[Vulcan]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

it would not have to be reported w/ AR, but certainly r12 will have this data and could easily auto-upload the info.

So you want to create an entirely sepperate database? I would think it would make more sense to tie it to AR.

[EliC]

Re: Par2Par: p2p PAR2 sharing for CD rip repair

So you want to create an entirely sepperate database? I would think it would make more sense to tie it to AR.

spoon just said AR doesnt have access to this data. Whatever is easier to implement. It doesnt even need to be a database, just a sufficient sample size.