Thread: SACD levels
View Single Post
  #25   Report Post  
Posted to rec.audio.high-end
Steven Sullivan
 
Posts: n/a
Default SACD levels

Stewart Pinkerton wrote:
On 15 Dec 2005 03:12:29 GMT, Steven Sullivan wrote:


chung wrote:
Steven Sullivan wrote:


snip


In the case of dynamic range compression, I don't understand how the
PCM format "allows" this "abuse" while DSD does not. In both cases the
mixing/mastering engineer is free to apply dynamic range compression.


This already appeared once in this thread:

Nika Aldrich said:
"
[...]
The CD is a very versatile format that actually allows illegal content -
waveforms that exceed its own boundaries. This can cause distortion upon
playback through D/A converters. The SACD has, built into its
specification, a system that prevents this from occurring - a sort of
mathematical algorithm that data passes through to analyze whether the
signal exceeds the boundaries of the converter. If it does, the signal
can't pass. The algorithm is actually fairly simple - 28 consecutive
samples.


This is a little confusing. I think he meant to say 28 consecutive
identical samples (all ones or all zeros).


They don't have to be ones or zeros to be identical.


Not sure he meant that -- rarely would you see 28 identical samples in something
that wasn't clipped.


That would be the point................


But sure, you really want to avoid 28 identical samples at
0 dBFS. The thing is, nowadays, lots of CD mastering uses the trick of reduing the
level to less than 0 db, *after* its already been clipped. THe result looks like flat-topped
.waves that never hit 0 dbFS.


And they would have 28 or more identical samples, representing a flat
top more than 0.6 milliseconds wide. It's not rocket science.........


I had in mind the fact that Cool Edit/Audition wouldn't recognize those 28 identical
samples as 'possible clipping', if they weren't at 0 dBFS. ( And also that all
samples are in some context 'consecutive', of course, so Nika indeed meant 'identical'
consecutive samples)

If SACD spec disallows 28 consecutive identical samples at levels *other than* 0 dBFS,
more power to it....that is the point on which I was unclear.
Though I gather there must be *some* intelligent discrimination
of level in the spec, otherwise digital silence would be illegal too.

--
-S
"The most appealing intuitive argument for atheism is the mindblowing stupidity of religious
fundamentalists." -- Ginger Yellow