Modelling the sensation of fluctuation strength

Abstract

The sensation of fluctuation strength (FS) is elicited by slow modulations of a sound, either in amplitude or frequency (typically <20 Hz), and is related to the perception of rhythm. In speech, such periodicities convey valuable information for intelligibility (prosody). In western music, most of the envelope periodicities are also found in that range. These are evidences of the potential relevance of FS in the perception of speech and music. There is, however, no published computational model to assess the FS of a sound. This might be one reason why when slow modulations of a sound are to be analysed, other indirect measures (e.g., loudness to estimate “loudness fluctuations”) or more complex techniques (e.g., the modulation filter bank) are used. In this paper a model of fluctuation strength is presented. The model was developed taking advantage of the physical similarity between FS and the psychoacoustical sensation of roughness. The FS model was then adjusted and fitted to existing experimental data ...