Predicting the Loudness of Non-Stationary Sounds: Zwicker's Original Envelope Extraction vs. DIN 45631/A1:2010

All currently standardized procedures for computationally predicting the loudness elicited by non-stationary sounds (under normal conditions in a typical subject) are based on the work of Zwicker and co-workers. The corresponding algorithms consequently share a spectral channel-based processing scheme, roughly mimicking peripheral auditory processing. With the same motivation, only the temporal envelopes in the channels are analyzed further. This contribution compares the envelope-extraction procedure proposed by Zwicker and co-workers to that of DIN 45631/A1:2010, regarding its intended purpose, implementation details, and effects on the predicted loudness. Based on the results, implications for future loudness-prediction standards are discussed.