Page 123 - ITU Journal, ICT Discoveries, Volume 3, No. 1, June 2020 Special issue: The future of video and immersive media

ITU Journal: ICT Discoveries, Vol. 3(1), June 2020

A NEW SET OF DIRECTIONAL WEIGHTS FOR ITU-R BS.1770 LOUDNESS
MEASUREMENT OF MULTICHANNEL AUDIO

1
2
3
2
Leandro Pires , Maurílio Vieira , Hani Yehia , Tim Brookes , Russell Mason 3
1
Agência Nacional de Telecomunicações, Belo Horizonte, Brazil
2
Universidade Federal de Minas Gerais, Belo Horizonte, Brazil
3
Institute of Sound Recording, University of Surrey, Guildford, UK
Abstract – The ITU-R BS.1770 multichannel loudness algorithm performs a sum of channel energies with weighting co-
efficients based on azimuth and elevation angles of arrival of the audio signal. In its current version, these coefficients
were estimated based on binaural summation gains and not on subjective directional loudness. Also, the algorithm lacks
◦
directional weights for wider elevation angles (|φ| ≥ 30 ). A listening test with broadband stimuli was conducted to collect
subjective data on directional effects. The results were used to calculate a new set of directional weights. A modified ver-
sion of the loudness algorithm with these estimated weights was tested against its benchmark using the collected data, and
using program material rendered to reproduction systems with different loudspeaker configurations. The modified algorithm
performed better than the benchmark, particularly with reproduction systems with more loudspeakers positioned out of the
horizontal plane.
Keywords – Broadcasting, loudness, signal processing, spatial audio, subjective test.

1. INTRODUCTION gles, the Rapporteur Group decided to keep simple posi-
tional weights envisioning backward compatible extension
In recent years, audio signal normalization based on loud- of ITU-R BS.1770-3 and, consequently, avoiding imple-
ness became a regular practice and the loudness algorithm mentation issues.
described in Recommendation ITU-R BS.1770 is now
ubiquitous in broadcasting and content producing work-
flows. With the advent of object and scene-based sound Table 1 – Position-dependent channel weightings in ITU-R
systems, it is reasonable to wonder if a channel-based BS.1770-4
loudness measurement, originally designed for stereo and Azimuth (θ)
Elevation (φ)
5.1 content, would still suit the needs of program level ◦ |θ| < 60 ◦ 60 ≤ |θ| ≤ 120 ◦ 120 < |θ| ≤ 180 ◦
◦
◦
management and control. After a hiatus, Rapporteur |φ| < 30 1.00 (± 0 dB) 1.41 (+ 1.50 dB) 1.00 (± 0 dB)
else 1.00 (± 0 dB)
Group on loudness measurement algorithm (RG-32) was
re-established in March 2019 to resume studies on mea-
surements for object and scene-based audio in light of the At that time, RG-32 conducted experiments with two dif-
new Recommendation ITU-R BS.2127, which specifies the ferent datasets using 5.1 and 22.1 loudspeaker configura-
reference renderer for these immersive audio formats [1]. tions. The obtained Pearson correlation coefficients, de-
Whether the algorithm will evolve to a format-based loud- scribing the linearity of the relationship between mea-
ness computation or remain multichannel, and be used in surements and subjective results, were r = [0.817,0.902],
the output of the renderer, is still to be determined. which were lower than the correlation obtained in the
original tests of ITU-R BS.1770 (r = 0.977) [2, p.17].
An attempt to address this issue was made in the latest ver- Also, authors in [3] tested this set of weights using 5.1
sion of ITU-R loudness model (BS.1770-4), in which the and 7.1 systems with and without screen loudspeakers.
positional weighting coefficients were extended to an unre- From a common database comprised of object and channel-
stricted number of channels [2]. The weighting coefficient based content, listening tests were made in three test
G i for an i-th position is derived from Table 1, which is a sites and the observed correlation coefficients were r =
generalization of the original 5.1 weighing scheme for the [0.962,0.820,0.898], being the first coefficient the only
horizontal plane, and no weighting is applied in broader one on par with the correlation from the original ITU-R
◦
elevation angles (|φ| ≥ 30 ). Although RG-32 considered tests. Thus, the question on how to account for directional
differences of directional loudness in wider elevation an- weighting in the ITU-R loudness model is still valid.

© International Telecommunication Union, 2020 101