Automated detection of errors and quality issues in audio-visual content

Abstract

This paper presents a demonstration of a technology for automated detection of errors and quality issues in audio-visual content. The extensible system, called AVInspector, consists of a signal processing core combined with detection modules for universally applicable audio-visual analysis. The detection modules include algorithms to assess blocking artefacts, picture freezes, clipping, dropouts and noise. The technology can be used for both a stand-alone application and as library for integration within professional products. Typical purposes are the automated observing and monitoring of audio-visual applications. This includes the automated analysis of audio and video material at ingest or within an archive, real-time observations of streaming services and broadcasts as well as the assessment of transcoding results for multi-platform playout.