IPB

Welcome Guest ( Log In | Register )

EAC... how perfect? How does it work?, Redundant, but more detailed questions
Radetzky
post Mar 10 2003, 01:55
Post #1





Group: Members
Posts: 95
Joined: 10-March 03
Member No.: 5399



I've been reading HydrogenAudio for a couple of weeks now and think
this place is a real gem for audio nuts. wink.gif

I am about ready to proceed to backup about 250 CDs of mine on a
RAID-1 setup. I plan to rip all the CDs using a LGE HL-DT-ST RW/DVD
GCC-4240N drive and EAC 0.9b4.

Though, before doing do, I would love to have your input regarding a
couple of points I have a hard time with. I hope you won't find these
questions redundant. I know some, or most, have already been asked
but I never was satisfied with the answers (*not* that I think anybody
_has_ to provide an answer I will be happy with...).

So here goes for the questions :

1. Is it true that (at minimum) in any of the Secure modes EAC will
read at least twice every thing it reads? I ask so because there
are no confirmation anywhere in the software that it is doing so.
It would be nice if during the rip process a little string saying
"currently reading each sector twice" or something would appear.

When, if ever, does EAC perform only one read of the audio?

2. Exactly what does it read twice? How big is the chunk of data that
is being read? 10x2352 bytes? 2352 bytes? 1 bit?

3. I know this has already been asked at digital-inn.de, but I don't
think I understood the answer fully. How sure can we be that EAC
really does detect and "correct" errors?

I mean, lets say there is a minor scratch on one of my CDs. My
drive reads through the scratch.. the returned data is, say 0x2431.
It reads again and the data is still 0x2431. EAC thinks this is
fine. But the scratch is done in such a way that the reflected
information is always the same, AND wrong. How could EAC deal with
this physical anomaly?

I remember Andre (the author of EAC) said something like it was
very unlikely that the same data could be read from a scratch or
spot (or whatever damage there is) on the CD. But how sure are we
of that?

4. If I intend to keep all the ripped music on my computer and not
duplicate it on, say, CD-R media, should I go nuts with the
offsets?

I did tests myself and I tried to inject a -10000 offset and then a
+10000 offset and it didn't mater at all sound wise. I checked
with a WAV editor (Audacity) and the difference (the time-offset)
was really minimal and only cutting or adding silence (haven't
found a CD that starts with music too soon).

If I want to extract all the tracks as one big WAV file, am I right
to say that the digital extraction is successful anyway, that the
offset only induces a time-shift (a very, very small one)? Or is
there a subtlety I didn't grasp?

5. A very important (to me) question. When EAC commands the drive to
read whatever amount it has to during its rip process, what error
correction is performed by the drive? Is it using the Reed Salomon
ECC error codes from the disc?

In DAE mode (... is it a 'mode' ?), if the drive detects a problem
on the disc (scratch, etc.) will it try to interpolate between two
points or will it just send the data back unmodified?

Asked differently, in DAE mode does the drive treat the audio data
as audio or as data?

Reading http://www.roxio.com/en/support/cdr/cderrors.html leads one
to believe there are many more problems to deal with that "only"
scratches and finger prints. The actual CD might have errors
injected during its pressing (!).

6. I have a HL-DT-ST RW/DVD GCC-4240N drive. It is a low-profile
combo (DVD + CD-RW) drive. When I perform a rip using EAC and this
drive, will it be of the same quality from a rip performed from,
say, a Plextor Ultraplex drive?

7. Last question... what is the ETA for the next release of EAC? smile.gif

Thanks for your time and patience ! :-)
Go to the top of the page
+Quote Post
 
Start new topic
Replies
Pio2001
post Mar 10 2003, 13:21
Post #2


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Interesting questions indeed B)

[quote=Radetzky,Mar 10 2003 - 03:55 AM]1. Is it true that (at minimum) in any of the Secure modes EAC will  read at least twice every thing it reads?  [/quote]
No !
[quote=Radetzky,Mar 10 2003 - 03:55 AM]When, if ever, does EAC perform only one read of the audio?[/quote]
As soon as you check the "drive is capable of C2" checkbox. And also if the drive caches and the "drive caches" checkbox is disabled. Note that some drives (I can remember two posts) are reported to cache by Feurio and not by EAC, which would lead the drive not to read twice even with EAC properly configured (according to its own feature detection i.e. no cache).
[quote=Radetzky,Mar 10 2003 - 03:55 AM]How big is the chunk of data that
   is being read?  10x2352 bytes?  2352 bytes?  1 bit?[/quote]
27 sectors = 63,504 bytes for error detection, then less for error correction, according to the size of the error detected, when the red lights turn on.
[quote=Radetzky,Mar 10 2003 - 03:55 AM]How sure can we be that EAC
   really does detect and "correct" errors?[/quote]
It depends on the drive for the C2 mode. 99 % seems an average value. But it varies from zero to 100 %. The reading twice mode (no C2) is more secure, so it can be trusted more than 99.9 % (that is the C2 accuracy of my Teac drive). Test and copy is even more secure : it can detect errors unnoticed in secure mode, non accurate, cache, no C2 mode.

http://www.cdrinfo.com performs C2 accuracy tests (available for recent reviews only).
My own C2 accuracy tests : http://perso.numericable.fr/~laguill2/dae/dae.htm
[quote=Radetzky,Mar 10 2003 - 03:55 AM]the returned data is, say 0x2431. It reads again and the data is still 0x2431.  [...]  the scratch is done in such a way that the reflected information is always the same, AND wrong.  How could EAC deal with  this physical anomaly?[/quote]
It is very rare that a scratch turns wrong just one byte. Such cases have been reported by BobHere ( http://www.digital-inn.de/showthread.php?t...?threadid=16578 ) and someone here (Tigre or Halcyon, I think), among about 1000 CDs ripped, IIRC.
Most of the time, a scratch turns wrong several tens of bytes (see the increments of errors in Nero CD Speed quality check). When the reading is done again, some bytes that were wrong are right, and some that were right are wrong, and the difference is detected : http://www.digital-inn.de/showthread.php?t...?threadid=15838
[quote=Radetzky,Mar 10 2003 - 03:55 AM]how sure are we
   of that?[/quote]
I'd say that the superiority of reading twice vs C2 has been tested about half a dozen times, and reading twice always came better than C2, that must have been at 99% accuracy. But remember it's before all a matter of drive, and of caching.
[quote=Radetzky,Mar 10 2003 - 03:55 AM]should I go nuts with the
   offsets?[/quote]
No
[quote=Radetzky,Mar 10 2003 - 03:55 AM]am I right
   to say that the digital extraction is successful anyway, that the
   offset only induces a time-shift (a very, very small one)?  [/quote]
Perfectly right, glad to see someone at last, who understands that offset correction has nothing to do with secureness smile.gif
[quote=Radetzky,Mar 10 2003 - 03:55 AM]When EAC commands the drive to
   read whatever amount it has to during its rip process, what error
   correction is performed by the drive?  Is it using the Reed Salomon
   ECC error codes from the disc?[/quote]
Yes, CIRC (error correction for compact disc) is performed by the drive, otherwise the audio couldn't even be decrypted.
But the exact error correction algorithms used can vary. I saw different ones used by different drives : http://perso.numericable.fr/~laguill2/dae/...set/chipset.htm
In fact, the drive switches between different error correction strategies so as to optimize error correction according to the state of the CD : http://www.digital-inn.de/showthread.php?t...18281#post59836
[quote=Radetzky,Mar 10 2003 - 03:55 AM]In DAE mode (... is it a 'mode' ?)[/quote]
Yes : http://forum.cdfreaks.com/showthread.php?s...&threadid=47958
[quote=Radetzky,Mar 10 2003 - 03:55 AM]if the drive detects a problem
  on the disc (scratch, etc.) will it try to interpolate between two
points or will it just send the data back unmodified?[/quote]
My drives all interpolate : http://perso.numericable.fr/~laguill2/dae/...terpolation.htm . I think older CD ROM drive didn't interpolate, I remember my Artec 34x returning standalone spikes in damaged aeras of some CDs, that don't appear in my current drives. I think it didn't interpolate and returned uncorrected data.
[quote=Radetzky,Mar 10 2003 - 03:55 AM]Asked differently, in DAE mode does the drive treat the audio data as audio or as data?[/quote]
Processing data is using the third level of error correction present on CD ROMs. Audio can't be treated as data because this extra error correction codes are not on the CD.
[quote=Radetzky,Mar 10 2003 - 03:55 AM]The actual CD might have errors
injected during its pressing (!).[/quote]
CIRC should deal with errors injected during the pressing the same way that it deals with errors produced during the playback.
[quote=Radetzky,Mar 10 2003 - 03:55 AM]When I perform a rip using EAC and this drive, will it be of the same quality from a rip performed from,
say, a Plextor Ultraplex drive?[/quote]
No, the quality directly depends on the drive.
The secureness, on the other hand, is the same in test and copy, or non accurate, cache, no C2 mode. Some rare drives (Yamaha CRW6416 is a good example) have not accurate stream, the others will all be equally secure in cache/No C2 mode. Many drives can be unsecure in no cache mode. All drives seem to perform differently in C2 mode.

Edit : added the order of magnitude of the amount of CDs ripped by BobHere to find wrong CRC matching, updated links, put back interpolation pictures online
Go to the top of the page
+Quote Post

Posts in this topic
- Radetzky   EAC... how perfect? How does it work?   Mar 10 2003, 01:55
- - Mr. Mulder   The ETA for the next version of EAC is the same as...   Mar 10 2003, 09:00
- - Pio2001   Interesting questions indeed B) No ! As ...   Mar 10 2003, 13:21
- - tigre   Great answer, Pio2001 - You should add this thread...   Mar 10 2003, 14:49
- - AtaqueEG   QUOTE (tigre @ Mar 10 2003 - 09:49 AM)Great a...   Mar 10 2003, 18:40
- - NumLOCK   QUOTE (Pio2001 @ Mar 10 2003 - 01:21 PM)It is...   Mar 10 2003, 19:16
- - BadReligionPR   QUOTE (NumLOCK @ Mar 10 2003 - 10:16 AM)I per...   Mar 10 2003, 19:58
- - ben   QUOTE (Pio2001 @ Mar 10 2003 - 12:21 PM)No, t...   Mar 10 2003, 21:52
- - Pio2001   QUOTE (NumLOCK @ Mar 10 2003 - 09:16 PM)I per...   Mar 10 2003, 22:25
- - Pio2001   QUOTE (AtaqueEG @ Mar 10 2003 - 08:40 PM)Seem...   Mar 10 2003, 22:30
- - Radetzky   Thanks Pio2001. Your answers sure shed light on m...   Mar 11 2003, 04:06
- - Mr. Mulder   QUOTE (Pio2001 @ Mar 10 2003 - 06:25 PM)[]try...   Mar 11 2003, 07:42
- - tigre   QUOTE (Radetzky @ Mar 10 2003 - 07:06 PM)Fina...   Mar 11 2003, 09:06
- - tigre   QUOTE (Mr. Mulder @ Mar 10 2003 - 10:42 PM)...   Mar 11 2003, 09:57
- - Mr. Mulder   Thanks tigre.   Mar 11 2003, 12:00
- - Patsoe   QUOTE (Radetzky @ Mar 11 2003 - 04:06 AM)But ...   Mar 11 2003, 13:14
- - Pio2001   QUOTE (Radetzky @ Mar 11 2003 - 06:06 AM)EAC ...   Mar 11 2003, 13:33
- - EmDub   QUOTE Finally, EAC detects the following features ...   Mar 11 2003, 17:02
- - Radetzky   Pio2001: Hmm.. so to my question "if I rip a...   Mar 12 2003, 02:17
- - Patsoe   Radetzky: you're somewhat repeating your same ...   Mar 12 2003, 10:51
- - Pio2001   QUOTE (Radetzky @ Mar 12 2003 - 04:17 AM)Also...   Mar 12 2003, 12:43
- - yourtallness   QUOTE What do you mean by quality? Quality as in t...   Mar 12 2003, 13:27
- - Patsoe   QUOTE (yourtallness @ Mar 12 2003 - 01:27 PM)...   Mar 12 2003, 13:48
- - yourtallness   Let me rephrase: in digital audio, a 1 that should...   Mar 12 2003, 14:04
- - KikeG   I doubt error interpolation can result in less bas...   Mar 12 2003, 14:40
- - Patsoe   QUOTE (yourtallness @ Mar 12 2003 - 02:04 PM)...   Mar 12 2003, 14:44
- - tigre   QUOTE (yourtallness @ Mar 12 2003 - 05:04 AM)...   Mar 12 2003, 14:50
- - Radetzky   I understand I've been somewhat redundant with...   Mar 13 2003, 02:01
- - Patsoe   Usually, deleting silence does the job, yes. You c...   Mar 13 2003, 02:43


Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 20th December 2014 - 23:17