1. The recent brouhaha about the Metallica recordings
(1000 fans signed a petition complaining about heavy audio compression on their
latest release) (see HERE
-webmaster) led me to finish a project I've been mulling over for a long
time, namely to devise a way of measuring dynamic range of recordings objectively
in a way that is also meaningful subjectively.
At high levels there is no problem -- the ear's response is
relatively flat, and unweighted (or perhaps "C" weighted, which has
a bit of low end rolloff) is appropriate. However at low levels the ear has less
sensitivity at low frequencies, the lower the level the greater the loss. The
"A" weighting curve is commonly used to take this into account, however
if you examine the Fletcher Munson contours of equal loudness you see it corresponds
to the 50 dB spl curve. Thus it is not appropriate for very low levels such as
ambient noise in a hall or room. One ends up measuring traffic noise or "room
rumble" which is not audible. (Ideally one should have a meter whose frequency
response is continously changing with level. What professional acousticians do
is measure noise in third octaves and compare with charts to determine the "noise
What I did was add a 6 dB per octave rolloff below 400 Hz.
This mimics the 30 dB FM contour pretty well. It is 65 dB down at 20 Hz compared
with the (inverted) FM curve which is -64. Armed with this I measured the Verdi
Requiem on the BAS CD. Using a fast, average responding meter I measured 70 dB
from the loudest passage (measured flat) to the quietest (measured with the modified
A filter). The ambient noise was 6 dB lower still, and the peak value was 8 dB
above the average, so the total range was 84 dB which is exactly 14 bits.
In Brad Meyer's notes he says they used a custom 1.5 to 1
dBX compressor built by Rene Jaeger since they were concerned their 14 bit system
was not sufficient (allowing for headroom and uncertainty of level setting), and
indeed they were correct.
I measured a commercial recording with very wide dynamics,
the Pines of Rome, with Maazel conducting, and came up with 64 dB for the music.
If the Verdi is played back with peaks of 105 dB, then the quietest passage would
be at 35, so the weighting at the 30 dB level is reasonable.
Compression is pervasive in all the media except classical
music CDs, now least I can put a number on it.
President, Boston Audio Society
email me HERE