>The results are displayed using CoolEdit's frequency
>analyzer. See the graphs at
>http://www.tomshardware.com/consumer/02q3/020712/2u4u-05.html
Note that this analysis uses a linear frequency scale, so it shows
mostly the high frequency detail, which is useful, but do note this,
as nowadays we're accustomed to looking almost exclusively at log
scaled frequency graphs.
-Dan
________________________________________________________________________
________________________________________________________________________
>From Tue Mar 8 18:22:30 2005
Message: 13
Date: Fri, 12 Jul 2002 16:58:25 -0700
From: Marvin Humphrey <>
Subject: Re: ATRAC compression
Walter Knapp:
> Good luck Klas. I've been saying this for years and still get the same
> misunderstandings back from people who did not listen. They have gotten
> so wrapped up in what they think happens that they never bother to just
> get one and see if they are right. One of the things I have watched is
> those that do seriously try it don't come back with these arguments again=
.
You know, I'm so much in agreement that I almost wish I hadn't spoken up.
My conclusion is the same as yours with regards to nature recording. I had
hoped to illustrate the point by showing the circumstances under which
perceptual encoding _does_ become a problem, and point out that those
circumstances are unlikely to be encountered by nature recordists. You
know, giving shape to a shadowy monster, showing that there's nothing to
fear so long as you don't stick your hand right in its mouth.
As a mastering engineer working on a wide variety of music, for a variety o=
f
end media, I have run into problems from time to time. Modern pop
recordings are subject to _very_ aggressive processing at the mastering
stage -- clients demand it. This processing interacts unpleasantly with
downstream perceptual encoding. For instance, I made a lot of karaoke CDs
which used MPEG 1 encoding. MPEG 1 is pretty primitive and unforgiving as
far as perceptual codecs go. The less aggressive the limiting, clipping,
and multi-band compression going into the encoder, the better the end
result. The more pure the original source in general, the better the end
result.
Cymbals and roaring electric guitars, both of which have very complex
spectral content, are among the hardest to make sound right. Also,
Electronica frequently freaks encoders, producing surprisingly bad results =
-
because it is often an unnatural signal that the codec designers didn't --
couldn't -- anticipate. These observations aren't mine alone, they have
often been made by people doing shootouts of MP3 encoders; the same
principles apply to ATRAC.
However, if you begin with natural sounds of reasonable level, that is
precisely what the encoders were designed to handle. They will perform at
their maximum capabilities -- which are pretty astonishing.
-- Marvin Humphrey
CD design website - http://marvin.mrtoads.com
________________________________________________________________________
________________________________________________________________________
|