Aaron Ximm:
> >> > So, I'm noticing that when I USB-transfer a standard MD to
>>> > my computer -- I'm assuming that the compressed MD is
>>> > reconstituted as a now, lossy AIFF file -- the file size is
>>> > around 737mb. How big will the file be from a HiMD disc?
Walt:
>It is incorrect to call it a lossy AIFF file, the loss occurred when the
>piece was recorded, both by sampling loss in the A/D and by the
>compression. There was also a small loss due to the error rate of the
>recording medium.
This is possible but not likely. The error correction systems for
reading optical discs do just that--correct the inevitable reading
errors to deliver a perfect file.
>There will be another change when the file is passed
>through the D/A so we can hear it.
Yes, since the data recording is lossless (in non-compressed
formats), the A/D and D/A are the sole sources of degradation in a
digital recording system. It surprises me that after over twenty
years of digital audio, we still don't have a standardized way for
evaluating and comparing the errors of converters.
> > To clarify, AIFF (and the PC equivalent WAV) is not lossy(!). These
>> files are full-resolution uncompressed renderings of either compressed
>> MD (or HiMD) files, or perfect copies of uncompressed PCM HiMD files.
>
>Well, not very lossy. When you start digging around in how programmers
>handle data transfers and processing those files, you often find
>resampling and such like going on. And we have talked about how all
>storage media has a error rate. These are generally tiny and
>insignificant changes, but not always.
I just want to reiterate that 99.999% of the time the storage medium
makes -no- change, not "tiny and insignificant changes." I did
ten-generation subtraction tests and listening tests when people were
believing (incorrectly) that DAT recordings had generation loss.
Another pet peeve of mine is that every professional error-correcting
digital audio system (CD player, for example) should have an
error-correction headroom meter so you know when the medium is close
to failure. None do. Pros want to know.
-Dan Dugan
|