View Single Post
Old 2008.09.01, 08:56 PM   #8
Glathannus
True Final Boss
 
Glathannus's Avatar
 
Join Date: Feb 2006
Location: United States
Posts: 2,423
Glathannus knows what you did last summerGlathannus knows what you did last summerGlathannus knows what you did last summer
Default

WavPack files are best created from an official commandline encoder behind a GUI. Doesn't mean you have to manually input the commands for each individual file, because the GUI will only ask you for the parameters once ever (during setup). If you tried ripping with a program like dBpowerAMP (extremely easy to setup), it won't offer you WavPack's 'extra' option. If you like giving your computer something to do, you can have WavPack's commandline encoder spend 6x the usual amount of processing, to reduce an overall album by a few more MBs - with no further impact on sound quality or decoding speed.

Use FLAC if you're impatient. There's really not much 'wrong' with it, and if you have a Rockboxed player (or one of the elite Cowons), FLAC is going to give you the most battery life. If you own a lot of CDs and constantly find yourself on the verge of running out on hard drive space, then WavPack is a significant help, and without cutting corners like Monkey's Audio does.

My rips take about a half hour before I can start listening to them. That's mostly compression-related, and that's with two fairly recent CPU cores, each compressing different tracks simultaneously! In this era of on-demand entertainment, the average person doesn't want to wait more than 3-5 minutes before they can start listening to their own rip, because they don't care what the sound quality is, or how efficient the compression is, or whether or not the ripping process was jumping to conclusions about the 1s and 0s the CD/DVD drive thought it was reading.

WavPack's commandline encoder isn't actually very difficult to setup with Exact Audio Copy, but Offset Correction (another configuration that all lossless enthusiasts should go through) is huge pain in the ass if you've never done it before, and it's independent of which format you happen to be using. Almost every disc drive is different about how they rip CDs, even if the drives are the same speed and the same manufacturer. If you had a first or second generation PlexWriter Premium (my favorite drive for CD ripping), I could walk you through the full setup in half a minute. Otherwise if you don't know anything about the drive(s) you own (plus not all drives are created equal when it comes to reliable ripping), and if I don't know anything either, and if you don't have any special Test Discs... my WavPack tutorial would basically be walking you through how to do consistent compression/extraction off of unreliable CD readings (which merely preserves the problem with most people's CD readings rather than fixes it). If you're gonna spend the half an hour on crazy compression, you'd have a certain moral obligation to ensure that whatever raw audio you're feeding into WavPack, was already perfectly-read to begin with.

For any onlookers of this thread who think I'm speaking a bunch of audiophile hyperbole - my checksums don't lie.

This is what OSCA looks like when different drives are calibrated properly:

EAC extraction logfile from 14. July 2007, 19:47 for CD
Tokyo Jihen / OSCA

Used drive  :         16X52X32X52COMBO   Adapter: 1  ID: 0
Read mode   : Secure with NO C2, accurate stream, disable cache
Read offset correction : 738
Overread into Lead-In and Lead-Out : Yes

Used output format : D:\Applications\Exact Audio Copy\WavPack\wavpack.exe   (User Defined Encoder)
                     320 kBit/s
                     Additional command line options : -hh -l -m -t -w "artist=%a" -w "tracknumber=%n" -w "title=%t" -w "album=%g" -w "genre=%m" -w "date=%y" -x6 %s %d

Other options      : 
    Fill up missing offset samples with silence : Yes
    Delete leading and trailing silent blocks : No
    Installed external ASPI interface


Track  1
     Filename O:\Music\Shiina Ringo & Tokyo Jihen\2007.07.11 - OSCA (CD) [Maxi-Single]\01. OSCA.wav

     Pre-gap length  0:00:02.00

     Peak level 95.9 %
     Track quality 100.0 %
     Test CRC CB47BAF1
     Copy CRC CB47BAF1
     Copy OK

Track  2
     Filename O:\Music\Shiina Ringo & Tokyo Jihen\2007.07.11 - OSCA (CD) [Maxi-Single]\02. Pinocchio.wav

     Pre-gap length  0:00:02.68

     Peak level 95.0 %
     Track quality 100.0 %
     Test CRC 3D1B3B7E
     Copy CRC 3D1B3B7E
     Copy OK

Track  3
     Filename O:\Music\Shiina Ringo & Tokyo Jihen\2007.07.11 - OSCA (CD) [Maxi-Single]\03. Kaban no Nakami.wav

     Pre-gap length  0:00:00.45

     Peak level 94.1 %
     Track quality 100.0 %
     Test CRC AE6A1FFD
     Copy CRC AE6A1FFD
     Copy OK

No errors occured


End of status report
EAC extraction logfile from 14. July 2007, 20:25 for CD
Tokyo Jihen / OSCA

Used drive  : PLEXTOR CD-R   PREMIUM   Adapter: 1  ID: 1
Read mode   : Secure with C2, accurate stream, NO disable cache
Read offset correction : 30
Overread into Lead-In and Lead-Out : Yes

Used output format : Internal WAV Routines
                     44.100 Hz; 16 Bit; Stereo

Other options      : 
    Fill up missing offset samples with silence : Yes
    Delete leading and trailing silent blocks : No
    Installed external ASPI interface


Track  1
     Filename O:\Temp\Test Rip\01. OSCA.wav

     Pre-gap length  0:00:02.00

     Peak level 95.9 %
     Track quality 100.0 %
     Test CRC CB47BAF1
     Copy CRC CB47BAF1
     Copy OK

Track  2
     Filename O:\Temp\Test Rip\02. Pinocchio.wav

     Pre-gap length  0:00:02.68

     Peak level 95.0 %
     Track quality 100.0 %
     Test CRC 3D1B3B7E
     Copy CRC 3D1B3B7E
     Copy OK

Track  3
     Filename O:\Temp\Test Rip\03. Kaban no Nakami.wav

     Pre-gap length  0:00:00.45

     Peak level 94.1 %
     Track quality 100.0 %
     Test CRC AE6A1FFD
     Copy CRC AE6A1FFD
     Copy OK

No errors occured


End of status report
This is also what my MD5s (of the WAVs) look like:

746a84dcae7b6fc32cc988da65316c58 *01. OSCA.wav
13d1101e145dde117c8fef0ea0b4d730 *02. Pinocchio.wav
4cf96a0a49fb2406bb287661220b3656 *03. Kaban no Nakami.wav
Now if I were to try ripping that same CD with the same drives uncalibrated, or the applied Read Offset swapped between those drives, I would end up with different CRCs, and different MD5s. Let's see what happens if I set my Plexwriter Premium to a Read Offset of 29 instead of 30...

EAC extraction logfile from 1. September 2008, 19:42 for CD
Tokyo Jihen / OSCA

Used drive  : PLEXTOR CD-R   PREMIUM   Adapter: 1  ID: 1
Read mode   : Secure with C2, accurate stream, NO disable cache
Read offset correction : 29
Overread into Lead-In and Lead-Out : Yes

Used output format : Internal WAV Routines
                     44.100 Hz; 16 Bit; Stereo

Other options      : 
    Fill up missing offset samples with silence : Yes
    Delete leading and trailing silent blocks : No
    Installed external ASPI interface


Track  1
     Filename O:\Temp\Test Rip\01. OSCA.wav

     Pre-gap length  0:00:02.00

     Peak level 95.9 %
     Track quality 100.0 %
     Test CRC F54EF7C0
     Copy CRC F54EF7C0
     Copy OK

Track  2
     Filename O:\Temp\Test Rip\02. Pinocchio.wav

     Pre-gap length  0:00:02.68

     Peak level 95.0 %
     Track quality 100.0 %
     Test CRC 88DB5BA2
     Copy CRC 88DB5BA2
     Copy OK

Track  3
     Filename O:\Temp\Test Rip\03. Kaban no Nakami.wav

     Pre-gap length  0:00:00.45

     Peak level 94.1 %
     Track quality 100.0 %
     Test CRC 12703BA4
     Copy CRC 12703BA4
     Copy OK

No errors occured


End of status report
b1bcbe5ff06d96ab4bd097a910bc0ffb *01. OSCA.wav
64e0bf18edca3ea6de1a7721db829504 *02. Pinocchio.wav
94dbbe659b22273fe4486a3a2dda5823 *03. Kaban no Nakami.wav
Now if you understand how checksums work, then you'd know that if checksums (especially independent MD5s of the WAVs) don't match, then neither do the files they represent! If two different CD drives and/or Offset values do not yield the same checksum, then they do not yield the same rip! Different drives left alone at a Read Offset of '0' (the default behavior of pretty much all rippers with or without the option), will commonly yield different checksums from the same CD than other drives left at '0', because they need different correction values.

It's like firing a gun that has kick to it - you can preemptively aim the gun in the opposite direction by an amount of degrees equal to the kick, knowing that when you pull the trigger, the gun will be on-target. Every gun model is different. The Read Offset value is basically your degree of counter-aim. That's how drives behave on CD ripping, at least with audio CDs - this problem doesn't happen with data CDs, because they contain the mapping information (beyond table-of-contents) that lets the drive know exactly where it's reading. The official Audio CD standard cut some corners on data integrity in order to fit more possible minutes onto a disc. Even if a particular disc doesn't use those extra possible minutes, it still has to conform to the official sloppy standard, or else the players probably won't read it.

You cannot just simply follow my example by punching in '738' or '30', in an attempt to get the same rip as me - not unless you had the same drive model(s), because almost every drive is different! Then even after you know your particular drive's Offset values, if your drive is not capable of overreading into Lead-In and Lead-Out (the vast majority of drives can't - most of the ones that can are older Plextors), there are a small handful of CDs you still won't be ripping properly, and your drive won't even know what it's missing, so neither will you!

This problem is not simply solved by using a ripper that reports nothing but the most obvious problems - that's eliminating the symptoms rather than the disease. You need a ripper that can not only puke (declare the most critical errors), but also cough and sneeze, otherwise you are basing the reliability of your rips (before lossless/lossy compression is applied), solely on the presence or absence of puking. That's a rather blunt and careless regard toward a process that has varying degrees of subtleties.

I implore any of you with the OSCA CD (especially first-press), to take your ripper-of-choice (whether it offers logs or not), and rip all three tracks the same manner you'd rip anything else, except rip them as WAVs this time. Then check the MD5s on those WAVs, using one of dozens/hundreds of MD5 scanners available for download (it's a standardized process - so different scanners will yield the same MD5s from the same files - you can go ahead and check that too). Chances are your MD5s won't match mine, and that means your rips aren't 100% the same as what is on the original disc and how.

I could try to use other discs as examples, but OSCA is probably the best example because it's new enough that everyone's CD of it will be the same (it's also one of the few discs I've bothered to rip with multiple drives already). Over the years the same album/single will be unofficially pressed differently than before, and then checksums won't match on the old pressing versus the new. It's not the most important thing to have the same checksums as someone else for the same general album or single - it is important to rip no more or less than what's actually there, on the particular pressing of the CD that you happen to have. Websites/software such as AccurateRip try to turn Offset Correction into a conformists movement, and that's not what I'm advocating here. It merely helps to have at least one CD that's the exact same as someone else who has already gone through a proper calibration. From a personal ripping standpoint (after you are somehow already calibrated) there is no 'right' or 'wrong' pressing to own even if they are different.

I will try to post logs/MD5s of Watashi to Houden if there's demand for it (because EMFers are more likely to own that than OSCA and have the exact same pressing as me), but the truth is I haven't even ripped mine yet.

And in the future I will TRY to post a full tutorial on Offset Calibration, but it's far trickier to explain than it is to actually implement, especially when everyone's drive is different and I probably don't know more about your specific drive than you do. Now that Coaster Factory is gone, I have some Sasquatch-size shoes to fill.
__________________
You know Tokyo Jihen is a supergroup, when you can't blame most of the members for wanting to pursue other projects.
Glathannus is offline   Reply With Quote