Path: utzoo!attcan!uunet!samsung!xylogics!world!burley
From: burley@world.std.com (James C Burley)
Newsgroups: comp.music
Subject: Re: dbx vs. Dolby C Noise Reduction
Message-ID: <BURLEY.90Jul17234142@world.std.com>
Date: 18 Jul 90 03:41:42 GMT
References: <590@lexicon.com>
Sender: burley@world.std.com (James C Burley)
Organization: The World
Lines: 50
In-Reply-To: fc@lexicon.com's message of 17 Jul 90 20:58:13 GMT

In article <590@lexicon.com> fc@lexicon.com (Frank Cunningham) writes:

   > 3. dbx having a "tendency to cut off leading edges of transient
   >    What causes this problem and how serious is it?
   I think this occurs when you are pushing the channel. The RMS / log
   control circuit allows the edge to pass uncompressed to the tape,
   which saturates. The expansion circuit then cannot restore the 
   original edge.

I'm not an audio techie, but when I used dbx for encoding cassettes of
wide-dynamic-range LPs (later CDs) of classical music, I got the impression
that the pumping was due primarily to a loss of highs (compared to the
rest of the spectrum) on the tapes.  No matter at what level I recorded,
for example, Telarc's recording of Carmina Burana, the opening number
clearly pumped as the cymbals crashed on "Statu variabilis".  My theory
at the time was that the compression (reduction of volume in this case) kicked
in at a level based primarily and initially on the volume of the
high-frequency cymbal crash and then primarily and subsequently on the
volume of the lower-frequency orchestra and chorus, as the cymbal crash
faded out (being struck).

On playback, if the highs ended up at a lower level on the tape, the instant
when the cymbal crash was the loudest thing would play back at a slightly
lower, and normally inaudible, level (without dbx).  But two things made
the pumping sound audible: dbx multiplied the error by a factor of 2 (since
it uses 2:1 compression or thereabouts); and the rest of the sounds, unlike
the cymbals, were non-percussive so the change from an artificially lowered
volume level immediately back to the normal level (as the cymbals faded and
the lower-frequency and more consistenly level orchestra and chorus became
the primary volume base) was much more noticeable.  In other words, it sounded
like the chorus and orchestra did a very quick forte-to-fortissimo step
right as the cymbal crash faded from its initial powerful sound (about
a quarter of a second, I'd say).

Piano music also had a consistent pump on it in the tapes I made for playing
in my car.  Strongly hit piano notes have a very sharp transient; perhaps
they too have, at the transient point, significant high-frequency content,
or perhaps they trigger the saturation Frank refers to.

Of course, I could be all wrong in my theory -- I've had personal
experience of the extremely high quality of Lexicon's audio engineering
staff, so I'm not about to suggest that I know more about anything audio
than a Lexicon employee!

My highest quality stereo is in my car, by the way, so it wasn't that system
that was causing the pumping -- it was clearly dbx.  Non-dbx recordings of
the same piece didn't have the pumping but had the expected large noise floor
(which was why I switched to using dbx), and needless to say, once I put a
CD player in the car, all those pumping problems went away.  Now the cymbal
crashes and the chorus sings straight out!