The aim of the study was to investigate whether South African free to air television and radio broadcast audio contains loudness fluctuations that fall outside previously determined limits of listener comfort. This is a relevant aim as consumers often complain about loudness fluctuations in broadcast audio (e.g. “why are the commercials so loud?”). Loudness is an inherently subjective phenomenon that is not only subject to differences in human perception from day to day, but also more specifically by the frequency content, localisation, spatialisation and duration of the audio stimulus. Traditional audio level meters only measure the audio signal or digital samples, and do not take any of these psychoacoustic phenomena into consideration. Broadcast audio has traditionally been regulated by specifying the permitted maximum level (PML) of the signal to avoid overloading the transmitter or over- modulation the broadcast signals. While this is necessary to keep the transmission inside the technical dynamic range of the medium, it does not correspond to the perceived loudness of these signals. With the addition of power dynamic range processing techniques, content producers and broadcasters were now able to raise the average level (and correspondingly the perceived loudness) without affecting the permitted maximum level or the peak level of the signal. Broadcasts were still compliant, but subjectively louder. As this process has not been done uniformly across various stations, and various types of audio, fluctuations occur both between stations, and between different segments on the same station. These fluctuations are the cause of listener complaints.
There has been a move in international regulators and broadcasters to make a paradigm shift from peak normalisation to loudness normalisation of broadcast audio content. Limited, to no adoption, of this new paradigm in South African broadcasting is evident. This study provides baseline data of the status quo of South African free to air broadcast audio to investigate whether it contains problematic fluctuations, and therefore whether a move from peak to loudness normalisation could possibly have a positive effect.
The study found that generally radio broadcasts suffered from greater and more problematic loudness fluctuations compared to television. Televisions broadcasts differed enough from station to station to cause inter-station loudness fluctuations outside previously determined limits for listener comfort, but not intra-station fluctuations. SABC 2 was found to be the loudest and SABC 3 the softest at this particular testing high-site. Radio broadcasts contained a large proportion of inter-station fluctuations, and while it varied considerably from station to station, each station contains some proportion of intra-station fluctuations. Advertisements were found not to be the loudness programme segment type. It was found that the following segments types were generally the softest to the loudest: talking, advertisements, links (interstitials) and music. Ikwekwezi FM was found to be the loudest station by far, with Lotus FM the softest, with a very wide difference of 13.7LU between their integrated station loudness values. The vast majority of broadcast audio was found to have a loudness range appropriate to the intended platform, but perhaps limited to the range appropriate the lowest common denominator, resulting in signals with high to extreme levels of dynamic compression and peak limiting.
The study also provided much greater levels of detail of the nature and extent of all loudness fluctuations, especially for radio loudness data. Additionally, ‘zap testing’ methodology was tested to simulate real-life inter-station fluctuation scenarios, and also found to be an efficient method for extrapolating overall station loudness for a larger set of stations.
Finally, the study recommends the paradigm shift from peak to loudness normalisation for all audio content producers and distributors in the country, and suggests the EBU R 128 recommendation as the most viable starting point.
Mini Dissertation (MMus)--University of Pretoria, 2016.