" stretching the octaves " is separate from " equal temperment "
With equal temperment the octaves are pure 2 X, but the other
intervals ( 2nd, 3rd, 4th, 5th, etc.) are compromised to facilitate
transposition.
Octave stretching is used to compensate for the inharmoncity of
the overtones of the vibrating medium of an instrument.
Some of us use the digital output. If you care to record the digital file, reinsert the CD and record it again and then compare the two files then we have data not opinion.
Do you have any idea of what the acceptable error rate is for CDs? In production they use an optical holographic scanner. Very fast and accurate way to find and count pinholes etc. (Had a nice chat with the folks who made the testers after an AES show. Actually a lot in common with a secret military gizmo.)
ES,
Pinholes in the polycarbonate or in the vapor deposition? And to what effect is this affecting the digital data stream? Are the pinholes large enough to create a read error? In other-words are these pinholes read as a 1 or 0 or to small to be read at the wavelength of the laser used to scan the disk?
Pinholes in the polycarbonate or in the vapor deposition? And to what effect is this affecting the digital data stream? Are the pinholes large enough to create a read error? In other-words are these pinholes read as a 1 or 0 or to small to be read at the wavelength of the laser used to scan the disk?
Pinholes are in the vapor deposited metalization layer.
They count pinholes, scratches and other flaws. Yes some of these cause bit errors that is why there is so much error correction.
I do have the test equipment to read CD errors. But anyone can do it. All you need to do is rip a CD wave file to your computers hard drive twice and compare the files.
ES,
I guess what you are saying is that with the optical reader there are definite read errors. I am trying to use the analogy of a data disk for something like a computer program where even 1 data error can render a program in error and why bit comparisons are done to make sure that the exact bit count is the same. Why is this any different for an audio disk and a data disk or Operating system disk?
I guess what you are saying is that with the optical reader there are definite read errors. I am trying to use the analogy of a data disk for something like a computer program where even 1 data error can render a program in error and why bit comparisons are done to make sure that the exact bit count is the same. Why is this any different for an audio disk and a data disk or Operating system disk?
Read errors are corrected to perfection by the error correction.
If the error (scratch or whartever) is too large, the reader tries to 'cover it up' by re-inserting a previous segment and this is normally quite audible.
Error correction operates 1000's of times on a typical CD and the data is non the worse for it.
The software CD uses the same concept but the error correction is more complex and extended because you can't even have one bit in eror.
jan
Pitch does not map to log frequency perfectly in the human auditory system. Things must be stretched a bit on the outskirts, and of course pitch perception practically disappears at the extremes.
I'd hoped to learn more about this and plunked down for a recent book* on the subject. Long and short of it: it's hairy and much is still not well-understood.
*Plack et al., Pitch, Neural coding and Perception, 2005, Springer, 0387234721
Error correction can only be perfect when the input/source is known. The optical reader does not know what it has missed or added that was or was not supposed to be there. Once what-ever is read, THEN it can be processed in an error-correct(ed) manner.
Tell me how we know for sure what was put/recorded onto the CD? With pin-holes and all the rest?
Thx-RNMarsh
Tell me how we know for sure what was put/recorded onto the CD? With pin-holes and all the rest?
Thx-RNMarsh
Last edited:
Another good book: MacWilliams and Sloane, The Theory of Error-Correcting Codes, 1977, North-Holland, 0444851933.
Error correction, by definition, is perfect.
When errors are sufficient enough that the error correction system can't perfectly correct and recover from the errors, it then resorts to error concealment, such as interpolation. And when it's too bad even for that, it will usually just mute the output.
se
When errors are sufficient enough that the error correction system can't perfectly correct and recover from the errors, it then resorts to error concealment, such as interpolation. And when it's too bad even for that, it will usually just mute the output.
se
BTW, examples of when this is not taken into account can be found in some electronic pieces, which presume that a chord or sequence can be indefinitely transposed up by precise frequency multiples of 2. The upper last versions sound woefully flat (e.g. Ashforth, "Byzantium", for electronic tape or organ and tape, which was produced on an early Moog synth.).
Better than eating it. I went to a listening session where the owners wife offered chocolate mousse made with instant pudding and cool whip, this was in my peak foodie days where that dish meant a trip to the farm for cream and the finest Belgian or Swiss chocolate.
Early CD players had an uncorrectable sample output, I could only get it to register by putting my finger on the edge of the disk.
Been there, the only time I had ONE bit difference was with a CD left on the floor of my car and ground into all the dirt there for an entire New England winter.
Steve,
I'll assume you didn't read this rather than didn't understand it. The article makes it clear that there are limits to the error correcting ability and how known uncorrected errors are masked.
Parity errors are easily detected for single bit errors. The longer your parity string the longer errors that can be detected. More complex is the issue of error correction. There are some errors than can be detected but the correction is not assured.
Now the "Modern" CD is still based on late 70's and very early 80's technology.
I was fixing a piece of data communications gear and found it was full of many familiar parts that I had used in some of my own designs. I mentioned how easy it was to fix to one of the manufacturers engineers. His response was "yeah it is full of 70's technology." I didn't correct him as the chips used were actually 80's parts. To this fellow who graduated in the 90's it was all ancient history. BTY he is one of the better engineers I know.
ES
Last edited:
Many of the decoder chips had a C2 error fail pin on them so you could monitor any instances where the error correction was not able to correct the error. Even on cheap players in the early days unless the disc was in pretty bad shape, you'd never see this pin go high.
se
- Status
- Not open for further replies.