118
lM6
and correlate well with human test panels. It therefore
seems justified to use Leq(RLB) as a baseline measure for
loudness, especially because room for improvement is also
built into the standard. The final BS.1770 standard included
a multichannel annex with a revised weighting filter, R2LB
– now known as “K” weighting - and a channel weighting
scheme. These two later additions have been less verified
than the basic Leq(RLB) frequency weighting.
The other aspect of BS.1770, the algorithm to measure
true-peak, is built on solid ground. Inconsistent peak
meter readings, unexpected overloads, distortion in
data reduced delivery and conversion etc. has been
extensively described, so in liaison with AES SC-02-01, an
over-sampled true-peak level measure was included with
BS.1770.
In conclusion, BS.1770 is an honorable attempt at
specifying loudness and peak level separately, instead
of the simplistic (sample peak) and mixed up measures
(quasi-peak) in use today. The loudness and peak level
measurement engine of LM6 follows the standard precisely.
Possible updates to the ITU standard may be released as
LM6 updates, provided that processing requirments doesn’t
exhaust the system.
Technical papers from AES, SMPTE, NAB and DAFX
conferences with more information about loudness
measurement, evaluation of loudness models, true-peak
detection, consequences of 0 dBFS+ signals etc., are
available from the TC website. Visit the Tech Library at
www.tcelectronic.com/techlibrary.asp for details.
Meter Calibration
Because of the frequency and channel weighting, and
of the way channels sum, only specific tones and input
channels should be used for calibration.
The most transparent results are obtained using a 1 kHz
sine tone for calibration. Other frequencies or types of
signal may be used (square wave, noise etc.), but don’t
expect similar results. The beauty of the system lies in
its RMS foundation, so this is a feature, not an error. The
same feature enables the loudness measure to identify
overly hot CDs or commercials, and to take out of phase
signals into account just as much as signals that are in
phase.
If we stick to standard methods for measuring peak audio
level in a digital system, where a sine wave (asynchronous
of the sample rate) with digital peaks at 0 dBFS, is
regarded a 0 dBFS tone, BS.1770 and LM6 output these
results:
One front channel fed with a –20 dBFS, 1 kHz sine tone =>
Reading of –23,0 LUFS.
Two front channels fed with a –20 dBFS, 1 kHz sine tone
=> Reading of –20,0 LUFS.
All 5.1 channels fed with a –20 dBFS, 1 kHz sine tone =>
Reading of –15,4 LUFS.
Level versus Loudness
When level normalization in audio distribution is based on a
peak level measure, it favors low dynamic range signatures
as shown in Fig 1. This is what has happened to CD.
Quasi-peak level meters have this effect. They tell little
about loudness, and also require a headroom in order
to stay clear of distortion. Using IEC 268-18 meters, the
headroom needed is typically 8-9 dB.
Sample based meters are also widely used, but tell even
less about loudness. Max sample detection is the general
rule in digital mixers and DAWs. The side effect of using
such a simplistic measure has become clear over the last
decade, and CD music production stands as a monument
over its deficiency. In numerous TC papers, it has been
demonstrated how sample based peak meters require a
headroom of at least 3 dB in order to prevent distortion and
listener fatigue.
The only type of standard level instrument that does not
display some sort of peak level is the VU meter. Though
developed for another era, this kind of meter is arguably
better at presenting an audio segment’s center of gravity.
However, a VU meter is not perceptually optimized, or ideal
for looking at audio with markedly different dynamic range
signatures.
Unlike electrical level, loudness is subjective, and listeners
weigh its most important factors - SPL, Frequency contents
and Duration - differently. In search of an “objective”
loudness measure, a certain Between Listener Variability
(BLV) and Within Listener Variability (WLV) must be
accepted, meaning that even loudness assessments by
the same person are only consistent to some extent,
and depends on the time of day, her mood etc. BLV
adds further to the blur, when sex, culture, age etc. are
introduced as variables.
Because of the variations, a generic loudness measure
is only meaningful when it is based on large subjective
reference tests and solid statistics. Together with McGill
University in Montreal, TC Electronic has undertaken
extensive loudness model investigation and evaluation.
The results denounce a couple of Leq measures, namely
A and M weighted, as generic loudness measures. In fact,
a quasi-peak meter showed better judgement of loudness
than Leq(A) or Leq(M). Even used just for speech, Leq(A)
is a poor pick, and it performs worse on music and effects.
An appropriate choice for a low complexity, generic
measurement algorithm, which works for listening levels
used domestically, has been known as Leq(RLB).
Combined loudness and peak level meters exist already,
for instance the ones from Dorroughs, but BS.1770 now
offers a standardized way of measuring these parameters.
In 2006, ITU-R Working Party 6J drafted a new loudness
and peak level measure, BS.1770, and the standard
has subsequently come into effect. It has been debated
if the loudness part is robust enough, because it will
obviously get exploited where possible. However, with a
variety of program material, Leq(RLB) has been verified in
independent studies to be a relatively accurate measure,
