ITU-R BS.1770 and EBU R128 were introduced to establish a measurement procedure that takes account of audible loudness differences between different broadcast audio material. The intention is to match the audio levels of 'loud' advertizing clips and the audio levels of regular movies or talk shows.
An open algorithm for the measurement of loudness and the true peak levels of programs with simple implementation are the basis of both standards. A so-called 'K-weighting' filter curve is defined. This filter curve is applied to all audio channels, except LFE. The goal is to achieve an objective measurement for a subjective impression of loudness.
The result of the 'K-weighting' filter curve is LKFS. LKFS is not a level, it is the loudness unit, the illustration of loudness referenced to the digital Full Scale. The single unit is 'LU', the Loudness Units. 1LU is equivalent to 1dB.
Controlling loudness the original content should not be changed. If signals are more or less compressed they should stay this way unchanged. The goal of loudness control is to change the overall level of different program parts in a way that these program parts sound equal (in perceived level) but are only changed in relative level to each other. Critical are here the time constants chosen within which levels are compared.
The LU measurement implements a multichannel loudness measurement algorithm. Usually the five main channels are used (left, centre, right, left surround and right surround). If less channels are used, these measurement channels would not be used. The LFE channel (low frequency effect) is not included in the measurement.
The algorithm implements two filters: a pre-filter with a high frequency increase (modelling the acoustic effects of the listener's head) and following high-pass filter (RBL).
|channel processing and summation|
left: pre-filter and high-pass filter (RBL)
right: combined filtering "K-weighting" compared to the
dB(A) weighting curve
|practical loudness control for similar received loudness|