As ever on this forum, a little knowledge.....
Data Is stored on a CD using a redundant self correcting encoding sceme known as a hamming code. The specific system is called Reed-Solomon encoding. Data is stored as blocks, and depending on the level of redundancy in the data, a given number of errors can be detected, and a given (smaller number) or errors corrected. The enthusiastic student should look up Reed - Solomon on Wikipedia.
The more redundancy in the data, the more robust the disk, but the less data can be stored. That's why yellow book CDroms (which have a higher data redundancy) store less data than redbook music CDs.
So, for a CD player doing a one pass 'on the fly ' read, the hardware error correction in the drive will attempt to correct any errors encountered. If the level of corruption is too high, it will be able to detect the error, but not fix it. In this case, the drive will either leave the value of the errored sample the same as the previous sample, or perform a linear interpolation to the next known good sample and make the missing data up (too old to remember which).
This is where a rip and a CD player differ. If the hardware error correction can't fix the error, a CD player has to move on, in contrast, if the controlling software of a rip receives an error flag from the drive, it can ask the drive to have another go. CD drives are mechanical systems, in particular the stepper drive to the laser head has backlash, and 'having another go' can place the laser in a slightly different position relative to the disc which allows a successful read to take place. This is why you sometimes hear a drive on a PC sawing away as repeated attempts to read an errored section are made by repositioning the laser read head.
So, it is quite possible for a rip to be able to extract more correct data from a cd than a CD player, doesn't make the music from the rip any better than the original recording, just less wrong than the CD player.
A note about the 'checksum'. This has nothing to do with the error correction on the drive, but is an overall test of the entire track calculated by the ripping software. It's a lot more complicated than Chris's example above - (google cyclic redundancy check) but it performs the same function. If two rips have the same checksum, then it is likely that the data is the same - this is a final test of accuracy of the rip, and if in error, some ripping program's will have another go at the entire track.
Thanks for that post. I didn't know that computer CD drive mechanisms could try rereading the data from different angles when necessary. I've learnt something new today already.
It doesn't re-read from a different angle, but shifting the mech back and forth will position the laser head in a slightly (nats EDITED) different position due to it being a mechanical system. This maybe enough to get a good read though.
Yes that makes sense.
PC > AVI Neutron Five 2.1
Sony NWZ-A847 64GB Walkman > Westone UM3x
Understanding the capabilities of digital audio, even if the method by which these capabilities are achieved is beyond you, as it is me, requires a bit of grounding in science and how science works.
Since science, particularly the science of digital audio, is quite clearly the work of the antichrist it is to be avoided at all costs.
We do so many shows in a row,
And these towns all look the same,
We just pass the time in our hotel room
And wander 'round backstage,
Till the lights come up, and we hear that crowd,
And we remember why we came.
My £10 PC CD reader can rip a disc, in ONE PASS at 40x speed with NO ERRORS! NONE. ZIP. ZERO ERRORS. Sorry to shout but some ppl just aren't getting the memo.
I'd love to see that CD drive that can read the inner edge of a CD at 40x speed with disintegrating to bits from the vibration.
(If you going to shout, please make sure that you are not shouting non-sense.)
Also from my experience, even with a new audio disc the experience you describe is rare. More often than not the rip will slow down at some point or other due to a relatively high level of read errors. For example due to the disc not being perfectly centered. In most case though, slowing down the read speed is enough to still achieve a 100% accurate rip. (Due to redundancy of the data.)
As far as I was aware data is stored on the CD as a series of 'pits' . These are read by the laser mechanism in either a CD player or in a CD-ROM drawer of a computer.
How exactly a 'read head' decides what is 'bad' data that shouldn't be there is something else altogether but if I had to trust something to read the data off an audio CD I know it wouldn't be attached to a computer.
A cd player that has to read all the data in one pass, irrespective of damage, misalignment, dust or other factors, then apply error correction circuitry to interpolate (informed guesswork) for missing data is, better than a computer drive that makes multiple passes, collects all the data and then performs a 'checksum' calculation to veryfy that the data is 100% accurate.
Just because the optical drive is in a cd player rather than a computer.
And how does that checksum know that the data is 100% accurate?? Oh I get it - AccurateRip. The truth is, if parts of a CD are scratched beyond repair, no amount of error correction is gonna recover the music, be it through software or otherwise.
Remember, audio on CD is stored as PCM, not WAV format, so to an extent comparing a rip to the CD is comparing apples to oranges. Some may shoot me, but things like AccurateRip are ultimately a red herring.
In all of this I'm not saying rips don't sound good. They do, and my personal preference audiowise is uncompressed WAV.
Arcam Solo Mini/Monitor Audio RX1/Cambridge Audio 751BD/Samsung 37” LCD
A checksum is not error correction, it is error detection - just a way to get confidence that the data is correct. The usual way to do this is to embed the checksum in the data iteslf - a real world example is your credit card number where a basic checksum is included to spot obvious mistakes and transposed numbers. If the checksum isn't embedded in the data, then it can be sent separately - but both these techniques require the checksum to be calculated from known good data before the data is sent.
In the case of a CD, there isn't a checksum available from known good data. Accuraterip (and similar programs) build a database of the checksums from users rips. When you rip a CD, your checksum is compared with other peoples checksum for the same CD. The chances that your rip is in error if (say) 30 people have ripped the same disk, got the same checksum as you are vanishingly small.
There is no red herring here - absent a checksum from known good data, having a common checksum from multiple separate rips is a good second best test.
As for different formats, that is a red herring. Data gets transposed losslessly from format to fomat all the time. As I have already posted, there isn't a bit for bit relationship with the data on a CD and the output of a CD drive - the error correction hardware has to strip out all of the additional error correction bits and reformat the data before it leaves the drive. When calculating a checksum, it does have have to be on the same data, but whether that data is in FLAC, OGG Vorbis, PCM or whatever doesn't matter.
So AccurateRip is reliant upon your CD being popular enough for it's database to hold multiple results from many others having ripped the same disk? (And the same version of the disk.)
Is there any way of searching for your CDs on their system before buying AccurateRip? (Or at least a trial version.)
Marantz M-CR603 + AirPlay • Rega R3 loudspeakers • iPhone 5 • iMac • Apple Airport Extreme 802.11n • Apple iPad Mini • Panasonic TX-L32D25B • Sony BDP-S390 • Ruark Audio R1 Deluxe • Humax HDR-Fox T2
I use dBPoweramp (which I would recommend) - it certainly shows how many rips there are of your CD when it has finished a rip, but I can't remember if it shows you how many are in the database before the rip. I have never ripped a CD where there wasn't at least two or three other rips of the CD already in the database - but maybe my tastes are very mainstream. I dont know if you can interrogate the system before you buy.
Absent an existing rip, dBPoweramp will do a multi pass rip - rip the CD twice and compare the checksum from the two rips. This is less useful as it will catch unstable errors (a bit was read as a 1 on the first rip, but a 0 on the second), but if the errors are stable (a 1 is read as a 0 both times), then the checksums will agree but will both be wrong.
I have never ripped a CD where there wasn't at least two or three other rips of the CD already in the database - but maybe my tastes are very mainstream.
Put it this way, Gracenote doesn't recognise the vast majority of my BBC Radio drama / comedy / documentary / history etc.. CDs (about 250 or more I think).
An example was the 9 CD set of The Diary of Samuel Pepys that I ripped last week. I had to paste in art work, type in artist, album title, volume numbers, disk numbers myself.
I sort of doubt much of my (non-music) collection is going to benefit from EAC, AccurateRip or DbPowerAmp or whatever.
I'm not missing them, I am entirely satisifed with the ALAC and 320K AAC rips (error correction always on) that I have already done. Just curious really.
you engineers and scientist..... I swear.
I know, I know......
Next time your car doesn't start in the morning, call a poet.
When your broken leg needs setting try a barman.
I am entirely satisifed with the ALAC and 320K AAC rips (error correction always on) that I have already done. Just curious really.
This is the best approach to ripping IMO - just use whichever ripping program you feel comfortable with.
I've tried EAC and dBPoweramp in the past but I'm just as happy to use iTunes or WMP too. A bit of knowledge and understanding of things can be useful and is interesting but people don't need to worry and obsess too much about the trivial details. Turn on error correction, select whichever file type and bit rate you want then just enjoy your music and don't worry.
You lot must have really been mistreating your CDs. I've ripped loads and hardly any had any errors at all. Those that did had one error. One error would effect what, a couple of seconds of a song? This topic seems to be obsessing over something that shouldn't really be a problem.
Round Things > Boxes > Wires > Boxes > Walls/Ceiling/Acoustic Stuff > Ears > Brain
Dave, don't get me wrong... I mean, I myself have phd's in engineer & scientist its just some of the stuff its clear some just ....lol..... lets move on.
The topic was about the quality of the copy.....and while an exact copy maybe possible a better copy than the orignal is impossible.
© 2013 Haymarket Publishing