Automatic loudness control in short-form content for broadcasting

Abstract

During the early years of the International Telecommunication Union (ITU) loudness calculation standard for sound broadcasting [ITU-R (2006), Rec. BS Series, 1770], the need for additional loudness descriptors to evaluate short-form content, such as commercials and live inserts, was identified. This work proposes a loudness control scheme to prevent loudness jumps, which can bother audiences. It employs short-form content audio detection and dynamic range processing methods for the maximum loudness level criteria. Detection is achieved by combining principal component analysis for dimensionality reduction and support vector machines for binary classification. Subsequent processing is based on short-term loudness integrators and Hilbert transformers. The performance was assessed using quality classification metrics and demonstrated through a loudness control example.

Ajuste do algoritmo de loudness ITU-R BS.1770 baseado em reverberação

“…Apesar da identificação de pontos de melhoria na regulamentação [3], [4], ela tem sido efetiva no controle da programação da radiodifusão digital terrestre. Contudo, o modelo multicanal de medida foi desenvolvido anteriormente ao advento dos novos formatos deáudio baseado em objetos e cenas para a Televisão em Ultra Alta Definição (UHDTV).…”