HydrogenAudio

CD-R and Audio Hardware => CD Hardware/Software => Topic started by: Joe Bloggs on 2002-10-23 12:27:33

Title: Deglitch.exe?
Post by: Joe Bloggs on 2002-10-23 12:27:33
deglitch.exe is bundled with EAC--do you just run it on a whole wav when a track can't be extractly correctly? Or can you specify which part to deglitch? (it seems quite likely that if told to search for glitches all over the wav the prog would mistakenly identify other parts of the song as glitches...  )

How DO you run it anyway?

Are there better free deglitch software out there?
Title: Deglitch.exe?
Post by: john33 on 2002-10-23 16:12:33
Quote
Are there better free deglitch software out there?

If you pick up this http://www.inf.ufpr.br/~rja00/files/deglitch.zip (http://www.inf.ufpr.br/~rja00/files/deglitch.zip) piece of software written by David Bryant, the author of WavPack, I believe it to be better than the EAC packaged software.
Title: Deglitch.exe?
Post by: Speek on 2002-10-23 19:18:19
Better yet, go here (http://home.wanadoo.nl/~w.speek/deglitch.htm) and pick up deglitch + front-end!
Title: Deglitch.exe?
Post by: Joe Bloggs on 2002-10-24 05:27:33
Re: ripping bad CDs...

Why is it that Burst mode is regarded as better than Fast mode? (IIRC)

Also, in future versions of EAC I would propose an intermediate mode between Secure and Fast mode where it can at least register the parts of the track that are bad so that they can be isolated for (maybe automatic) glitch removal, just like a realtime CD player does. Basically a Secure mode with zero retries. Would this be a good thing you think?
Title: Deglitch.exe?
Post by: Joe Bloggs on 2002-10-24 06:07:16
Well... tried David's deglitch...

It seemed to pick up a lot of 'glitches' that I don't hear as such...

Well, when a file sounded obviously glitched to me, deglitch agreed, with over 3000 glitches detected... but they're still there after deglitch 'corrected' them.  I guess I shouldn't be expecting too much... but the real mystery is that this track played flawlessly to my ears on a discman...     

I guess after-the-fact correction is simply missing some data that could have been used to correct the file during extraction... would 'Use C2 error information for error correction' be the thing?

I say, we need a 'CD player' extraction mode--a mode that can run at at least 1x on any track, however damaged, and concentrate on 'error hiding' using on-the-fly available information rather than trying to get that impossible 'perfect rip'.
Title: Deglitch.exe?
Post by: Pio2001 on 2002-10-24 11:33:48
Quote
I say, we need a 'CD player' extraction mode--a mode that can run at at least 1x on any track, however damaged, and concentrate on 'error hiding' using on-the-fly available information rather than trying to get that impossible 'perfect rip'.


I'm stuck in the analysis of the DAE quality of my drives, but so far I discovered that CD ROM drives (Memorex DVD Maxx 1648, Sony DDU1621, Yamaha CRW3200, Teac E540) also corrects the data on the fly if there are read errors. The process is a little bit less accurate than with a hifi player.
But I've especially noticed that the Teac E540 managed to correct more errors than the Yamaha Hifi player ! So it a matter of drive, before error concealment.
I don't know when I'll be able to publish the results of the analysis.


Quote
Why is it that Burst mode is regarded as better than Fast mode? (IIRC)

Because it reads without repositioning the head, that increases the risk of loosing the track when there are scratches.
Title: Deglitch.exe?
Post by: Joe Bloggs on 2002-10-24 13:32:59
So I suppose the CD-ROM drive's error correction is already kicking in when you set it to read in burst mode?

Do you know how to set the reading speed of the CD-ROM lower in win xp? (edit: or indeed, windows anything? I thought I knew how to set it in win9x but I just tried... and  is this dependent on the drive whether you can do this or not?) More importantly, do you think that would help with the error concealment quality?

I remember reading something to the effect that a Plextor set at 4x has much better error correction AND concealment (not sure about the concealment part) than the same drive set at max. speed. Also read in the same article that one drive can have worse error correction but better error concealment than another drive, etc.
Title: Deglitch.exe?
Post by: bryant on 2002-10-24 17:41:52
Quote
Well, when a file sounded obviously glitched to me, deglitch agreed, with over 3000 glitches detected... but they're still there after deglitch 'corrected' them.  I guess I shouldn't be expecting too much... but the real mystery is that this track played flawlessly to my ears on a discman...   :blink:  :o

I guess after-the-fact correction is simply missing some data that could have been used to correct the file during extraction... would 'Use C2 error information for error correction' be the thing?

Clicks in rips that come from damaged (or copy-protected) CDs come in certain patterns that are related to the interleaved method that data is written to a CD. DeGlitch is designed to detect and correct errors that fall into these exact patterns and avoid triggering on "wacky" samples that do not match the pattern. In these cases, DeGlitch can remove every audible glitch while leaving the audio perfectly intact, like in this sample (that EAC's deglitch function fails quite badly on):

www.wavpack.com/paul.exe (http://www.wavpack.com/paul.exe)    (WavPack self-extracting executable)

There are two situations I have found where it does not work. The first is where more than 1/3 of the samples are corrupt in a given burst. I have received a couple samples like this and have some ideas on how to improve DeGlitch to handle these cases, but haven't had the time.

Recently I have found another situation that is more difficult to handle. I have a Philips drive that is obviously trying to correct the glitches itself (probably in firmware), but unfortunately it is having trouble and in some cases making more of a mess than it started with. I can see the samples that are exact interpolations of the samples on either side (which is how concealment is implemented), but it acted on the wrong samples! After this botched operation, DeGlitch ignores the errors because they do not match the patterns for uncorrected errors. For example, if DeGlitch sees two consecutive "wacky" samples it knows that this cannot be a real glitch because they never come in consecutive samples (because of the interleaving). But, this is exactly what I saw in the output of the Philips.

A quick look in CoolEdit should tell you what is happening. It might be possible to adjust settings in EAC to get a rip that DeGlitch can handle better (to make sure that EAC is not trying to correct anything), or use another drive. In some cases it might be that the EAC deglitcher is more effective than DeGlitch (because it's not so picky), and in some cases the only solution would be to edit the glitches by hand or use a digital rip from the output of a CD player.
Title: Deglitch.exe?
Post by: Joe Bloggs on 2002-10-24 18:15:31
Thanks for the suggestions
Title: Deglitch.exe?
Post by: Mr. Crankypants on 2002-10-24 20:10:15
Wait, if you tried "deglitching" the average Autechre song, wouldn't there be nothing left?

Title: Deglitch.exe?
Post by: Pio2001 on 2002-10-25 22:37:42
Quote
So I suppose the CD-ROM drive's error correction is already kicking in when you set it to read in burst mode?

Yes
Quote
Do you know how to set the reading speed of the CD-ROM lower in win xp? (edit: or indeed, windows anything? I thought I knew how to set it in win9x but I just tried... and  is this dependent on the drive whether you can do this or not?)

EAC can set itself the extraction speed. If it doesn't work, some programs specific for the drives can do it. For example for the Teac E540. I didn't know the OS could do it too.

Quote
More importantly, do you think that would help with the error concealment quality?
I remember reading something to the effect that a Plextor set at 4x has much better error correction AND concealment (not sure about the concealment part) than the same drive set at max. speed. Also read in the same article that one drive can have worse error correction but better error concealment than another drive, etc.


It depends on the drive. The Plextor test is here http://www.exactaudiocopy.de/eac13.html (http://www.exactaudiocopy.de/eac13.html)

Read it anyway in order to understand the pictures linked below.

I ran the same test with the Teac E540, firmware 1.0a : slow speed, firmware 3.0a, high speed. However, the speed is under 1x as soon a s the CD is damaged.
The results are quite the same.

But with the Memorex DVDMaxx 1648, I tested audio extraction at max speed vs SPDIF audio playback at 1x, and the result is worse at 1x.

Results here : http://pageperso.aol.fr/lyonpio2001/dae/as...l/astraldae.htm (http://pageperso.aol.fr/lyonpio2001/dae/astral/astraldae.htm)

For the Teac, they are quite opposed to the results with the EAC test CD : http://www.digital-inn.de/showthread.php?t...?threadid=15244 (http://www.digital-inn.de/showthread.php?threadid=15244)

The main difference between the Cds is that the EAC test CD has a localised black mark causing a burst error, while the old CDR has little errors scattered everywhere on the CDR surface.