IBC2023 Tech Papers: NGA live-production workflows: Personalised audio experience in broadcasting

Abstract

Next generation audio (NGA) delivers personalised audio experiences on the basis of the listener’s preference and loudspeaker configurations and rendering algorithms in the listening environments. The prototype of NGA production equipment was arranged to verify the NGA live-production workflow to produce the typical personalised NGA programmes. It consisted of our developed NGA console, NGA loudness meter, and MPEG-H 3DA- complied coding system supporting the Audio Definition Model (ADM) and the serial representation of ADM (S-ADM). Subjective tests were conducted to verify the reasonability of the loudness measurement method in order to develop the NGA loudness meter using several rendering conditions, including loudspeaker configurations and rendering algorithms. The results showed that differences between objective and subjective loudness of NGA programmes are comparable to those of channel-based audio programmes.

Introduction

Next generation audio (NGA) delivers personalised audio experiences on the basis of the listener’s preference, reproduction setup, and listening environment. These experiences include users’ emphasis on dialogue and selection of preset programmes such as programme audio with a native and multilingual broadcasting content. This scenario is enabled by rendering processes that convert a set of audio signals with associated audio metadata into a different configuration of audio signals fed to loudspeakers. Audio metadata is used to describe the positions and gains of audio objects and the programme composition. As a world standard of audio metadata, the Audio Definition Model (ADM) is specified in Recommendation ITU-R BS.2076 [1] for offline production and the serial representation of ADM (S-ADM) is specified in Recommendation ITU-R BS.2125 [2] for live production.

NGA has been adopted as broadcasting standards (e.g., in Europe and the United States) [3 – 5] and its services have already been launched. In these services, offline-produced contents are mainly provided to users. On the other hand, live contents are rarely offered, and live-production workflows are not established. Case studies of NGA live-production workflows using S-ADM were reported by the EBU [6]. The reported workflows required the handling of several protocols and guaranteed synchronization between audio signals and associated S-ADM frames.

The authors have developed the NGA console [7], NGA loudness meter [8], and coding system [9] supporting S-ADM to realize NGA live production. This paper presents the NGA live-production workflow for a typical personalised audio programme and then an experiment conducted to validate the proposed workflow. Throughout the experiment, the production system consistently signalled S-ADM, which was output synchronized with audio signals from the NGA console. Within this workflow, the NGA loudness meter adopted the proposed method for measuring the loudness of signals obtained after rendering processes. To verify the proposed measurement method, the subjective test, which included loudness matching tasks, was conducted.