Digital Audio Extraction FAQ
What is Digital Audio Extraction (DAE)?
DAE is the process of extracting the audio from an audio CD by using a PC with a CDROM drive.
Is there such thing as a perfect DAE?
Yes and no. There are some drive/computer/software combinations that can extract an audio track bit for bit identical to the data on the CD. There are a lot of combinations that cannot. For example, using a Plextor drive and good software will almost always give perfect extractions.
How do you make sure wav file is accurate? How do you know you got a good extraction?
The best way to know if your wav file is good is to compare it to the data that was used to make the CD. In most cases this is not possible, so most people are left with extracting the track a number of times and comparing the files to see if they are the same. Usually a good tool for doing the compare is the DOS program FC. You may find that there are differences between two files that are both good extractions at the very start and ends of the file. This is caused by the initial seeking inaccuracy of the drive.
What error correction is on an audio CD? a data CD?
There is quite a lot of error correction built into every CD. For a better description than we could ever write, have a look at the following sites:
http://www.geffers.demon.co.uk/cd.html
What does a drive do when it finds an error?
It depends on how big the error is. For a small number of errors, the built-in error correction codes are used to fix the errors. For larger numbers of errors, a number of things can happen. Some drives take a guess (interpolate) on what the values should be based on the previous and following values. Other drives just set the bytes to 0. Some drives do a good job of reporting these conditions, while some others don't report any errors. It is all up to the person who programmed the firmware in the drive. You can see how many drives can give lots of errors in the final wav file but not report any errors to the DAE application.
Do drives do error correction when doing DAE at high speed?
There has been much discussion on this subject so the answer is difficult. The easy part of the answer is that there are fundamental parts of the data coding that force the drive to do error correction regardless of what the extraction speed is. As for drives doing interpolation, and other data correction, it is purely up to the firmware as to what is done and not done. Many drives simply don't have enough computing power to do the same level or correction when extracting high speed.
Reasons why wav files don't compare 2 times in a row
- crappy drives
- poor/no jitter correction
- bad wav headers
- bad parsing of header
- long blocks of digital silence confusing compare
- oak 910/911 chipset bugs
- bad firmware
- digitally generated tones, perfectly repeating confusing jitter correction routines
- normalizing of tracks
- frame slip
- stereo channel reversal
- upper/lower byte reversal
- ATAPI drive on same controller as old slow IDE drive
What is this jitter thing? or "It's not jitter!"
Again this is a question that is best answered by others instead of us:
http://www.fadden.com/cdrfaq/faq02.html#[2-15]
http://www.digido.com
What makes Plextor drives so good?
Plextor has always been known for making great (and expensive) drives. For those doing DAE, these drives are second to none. Why is that? The key to great DAE is being able to accurately read the audio tracks without losing your place. Plextor does this by adding some extra circuitry to their drives that is used to generate sync signals when reading audio discs. Once you have an accurate sync signal, it makes the process of accurately positioning very easy. When a drive can accurately seek to a specific point on the CD over and over again, it makes it very easy for the software to read the audio data and not have to do jitter correction and other fixes.
IDE (really ATAPI) vs SCSI?
Again this is a topic of religion. There are those that believe that SCSI is the only solution and there are others that believe that ATAPI works fine. In our experience, we've seen good ATAPI and SCSI drives as well as very bad ATAPI and SCSI drives. This means that the bus interface doesn't make the drive a good drive. However, we have seen that in general most SCSI drives do a pretty good job of DAE while most ATAPI drives don't.
Others will discuss the merits of SCSI over ATAPI based on the amount of CPU time that is needed to run the drive. We've found that SCSI does come out on top in this category especially when the ATAPI drive is using PIO drivers instead of BusMaster drivers. On the other hand, there are so many versions of BM drivers that are crap thus forcing the user to use PIO drivers.
CRCs in ripper programs what do they really tell?
A number of ripper programs out there will display a CRC value at the end of the track. CRCs are a method of applying a mathematical formula to a block of data to give a resulting value which is very likely different if even a tiny piece of that block of data changes the next time the CRC is calculated. What you are going to find is that if your drive/drivers/computer/software combination can rip a track a number of times in a row and give you the same CRC values, then you can be pretty sure that you are getting very good results from your setup.
What causes pops and clicks in wav file?
Again, the CDR faq has a much better description of this than wer could ever do:
http://www.fadden.com/cdrfaq/faq03.html#[3-3]
What causes pitch change?
Many people find their resulting wav file suffers from a change in pitch. Apart from the problems of a system that isn't generating a proper clock for playback, there can be a couple of reasons that the wav file itself could be at fault. If the drive is suffering from frame slip, or sample slip, the overall wav file will end up containing either fewer or more samples than it was supposed to and this will cause a change in pitch. It will also likely introduce pops and clicks.
Why don't my cdr copies sound like the originals?
We don't want to get into the religious war of peoples hearing, this is purely a bits and bytes discussion of can you get an exact copy of the data.
http://www.fadden.com/cdrfaq/faq04.html#[4-18]
How do we know you can really get perfect DAE?
We did a simple and unscientific experiment. What we found was that it IS possible to make a perfect copy of a track as far as the bits go. Effectively we did the following two experiments:
wav file->pressed CD->DAE->wav file compare
wav file->burned CD->DAE->wav file compare
Here's a site with another experiment on DAE quality:
http://www.treworgy.com/cdr/test.html
Equipment used:
Plextor 8Plex SCSI cdrom drive
Panasonic 585 ATAPI cdrom drive
CDDA32 DAE software version 1.1
AudioGrabber DAE software
Yamaha CDR-102 CDR drive
Maxell CD-R74H blank CDs
Test Preparation
We put together a wav file of some high quality audio which had been digitized in a studio. We then used a CD pressed by Quebecor at their manufacturing plant that used the wav file to create an audio track at the end of our data CD. We also took the wav file and burned a CD using our Yamaha 102 drive. This means we have a perfect original wav file to compare against the ones we make during the experiments.
Let's call the manufactured CD disc 'A', and the burned CD we'll call disc 'B'. First, we carefully cleaned the CDs before doing any tests. Next we extracted the tracks into wav files first from manufactured CD and repeated for burned CD. We extracted each track 5 times from both the Plextor drive and from the Panasonic drive using CDDA32. We repeated the extractions using AudioGrabber.
What we found
We compared the original wav file to all the wav files extracted from disc 'A' and from disc 'B'. Because the original wav file was configured with a fairly long period of digital silence at the start and end of the track, We configured our comparison tool to ignore digital silence at beginning and end of track and only compare the actual audio data in between. This means that initial seeking error at the start of the track wouldn't affect the testing if the body of the audio data was correct.
We found that ALL the wav files from the Plextor drive matched exactly. We found that some of the files from the Panasonic drive matched. The wav files that didn't match had either a whole frame added or lost occasionally. It seems our Panasonic drive has periodic problems with frame slipping. We repeated the tests on the Panasonic drive using jitter correction turned on in the software and found the wav files matched exactly.
As an aside test, we repeated a few of the tests on a Mitsumi 8x drive and we did not get exact comparisons. Of course Mitsumi has never been known for making good drives for DAE.
What can we conclude?
Yes Virginia, there is such thing as a perfect DAE, but you need the right combination of hardware and software to do it. We also know that it is possible to get good DAE from both a pressed CD and a burned CDR.
