What is bit perfect?

In Audioasylum PCAudio board, some contributors go on and on about hard disk outputted audio streams as being bit perfect and therefore much better than cd audio.

What is bit perfect what what do people here think. There has to be hard and software controlling all this and my take is that this is just a black box approach to the issue.
 
gmarsh said:
bit perfect = no bit errors.

If you keep your CDs in good shape and your CD player tracks well, you'll have "bit perfect" output.
--------------------------------------------------------------------------------

I know this. However, in practice how is bit perefct measured precisely. Is it true that HDs are much better than dedicated audio CD drives?

There are two parameters, time and 0 or 1. Does jitter then mean that no audio stream can be bit perefct on a time and value basis? Does bit perfect mean 'perefct audio stream fed to a dac? (Just like perefct 14 bit dacs in the beginning!)
 
No one authoritative seems to be responding, so I will, hoping that your thread will stay on the screen long enough for more comment.

The bit perfect programs read a segment of the disk multiple times and compares them until it comes up with a pretty good statistical indication that it has gotten the info off correctly, then moves on. This takes more time than just reading it once. Enough more time that it wouldn't work for real time playback. But there are very few errors.

As far as I've heard, a CD recording doesn't have all the checks and balances that a CD-ROM (data) or HD use to ensure almost perfect reading of data- its a lot sloppier, and NOT a perfect digital "1" and "0" storage.


If the CD player finds a section that it can't read well, it will fill in the gap with data "error corrected data " created using what the surrounding data is like. However media such as CD-ROM data disk or a hardrive pretty much have to replay things perfectly or the programs don't work very well........

So: the CD is not inherently an error free medium, the bit perfect programs take their time reading the disk, checking and rechecking and put it on a medium that IS checked for accuracy and does pretty much replay EXACTLY what is put onto it.
 
Variac said:
No one authoritative seems to be responding, so I will, hoping that your thread will stay on the screen long enough for more comment.

The bit perfect programs read a segment of the disk multiple times and compares them until it comes up with a pretty good statistical indication that it has gotten the info off correctly, then moves on. This takes more time than just reading it once. Enough more time that it wouldn't work for real time playback. But there are very few errors.

As far as I've heard, a CD recording doesn't have all the checks and balances that a CD-ROM (data) or HD use to ensure almost perfect reading of data- its a lot sloppier, and NOT a perfect digital "1" and "0" storage.


If the CD player finds a section that it can't read well, it will fill in the gap with data "error corrected data " created using what the surrounding data is like. However media such as CD-ROM data disk or a hardrive pretty much have to replay things perfectly or the programs don't work very well........

So: the CD is not inherently an error free medium, the bit perfect programs take their time reading the disk, checking and rechecking and put it on a medium that IS checked for accuracy and does pretty much replay EXACTLY what is put onto it.

As I understand it, the error correction on a CD corrects for read errors in such a way that the result is perfect. So, you can read the data again and again, but you always get the same data, and in fact you don't know if there was error correction or not, it's transparent.
In the case the error correction cannot correct the errors, error concealment takes over and start to interpolate and fill in the gaps a s best it can. If it gets any worse, the output is muted.
But reading again and again to find the "correct" data is useless; it is the same all the time (barring really catastrophic errors as described above).

Jan Didden
 
My computer drive slows waaaaay down when trying to extract a difficult CD ie .2x . It seems to me that there must be some correlation between speed and accuracy, that reading the data slowly is more accurate. I agree that reading the exact data over and over couldn't help, but my impression is that each reading comes out slightly differently- in which case rereading enough times to establish the most accurate read would be helpful.

Possibly this all only applies when there are areas that require error correction. When error correction is required, then the extraction could have advantages.

Possibly there is a lot more error correction going on than we would like to think,even on disks in good condition, and making error correction better would indeed improve playback
 
Just yesterday I edited a recording of a concert with GOLDWAVE, saved the .wav files, put it on audio CD with NERO. Then I ripped it again back to a .wav file with GOLDWAVE. Then I read extracted both the original and the CD-version with MATLAB, and compared them: perfect match, no bit errors.

So, a cheapy cd-rom recorder has allready bit-perfect performance?
 
dhaen said:
Jan,

I guessed what you were talking about, but couldn't resist the tease:angel:
Of course it's possible to error correct when copying, but sometimes it's better to copy the best "as is" data and error conceal far downstream.


Well, the Redbook Spec says it's not your business when to correct errors. A conforming drive does that continuously (because there are continuosly, thousands of bit errors on the disk) and transparantly. Whether you like it or not, the data comes off the disk essentially error-less. So there is no "as best" data; it's all the same, every read.

Jan Didden
 
So there is no "as best" data; it's all the same, every read.
Jan,

This just can't be so. If a dust particle shadows a "pit" on one read, and not on the next, then the data is different.

Now we could argue about bit-correction. There will be correctable errors, and uncorrectable errors. It's also up to us when we error correct. I kow the Redbook is for player specification and standardisation, but the reading of raw data from (Red Book compliant) discs is surely beyond the scope of Redbook.
 
Variac said:
My computer drive slows waaaaay down when trying to extract a difficult CD ie .2x . It seems to me that there must be some correlation between speed and accuracy, that reading the data slowly is more accurate. I agree that reading the exact data over and over couldn't help, but my impression is that each reading comes out slightly differently- in which case rereading enough times to establish the most accurate read would be helpful.

--------------------------------------------------------------------------------This is what I have been trying to explain to the flat Earthers in Audio Asylum! Bit perfect is only the result of a statistical or averaging model
that is as valid or invalid as error correction or interpolation. It doesn't mean perefct audio streams as some seem to be saying.

What do people think?

Another fallacy seems to be the assumption in the same forum that packet audio transfer like usb is free of jitter, regardless of hardware, software and method of clocking.

:smash: