Natural sounds are identified by their statistical properties, not by predictable patterns of elements
Evidence
For many natural sounds, such as wind, fire, rain, water or bubbling, integration of statistical information under continuous conditions is critical to assess the predictability of changes. While their spectrogram is dominated by small, recognizable elements (e.g. gust, crackles, drops), their occurrence is not predictable exactly , but they occur randomly in time and frequency, yet constrained by certain probabilities. As shown previously, these probabilities/statistics define the identity of such sounds, so called auditory textures (McDermott et al. 2013 ). McDermott and others could demonstrate this point directly, by recreating auditory textures from samples, using only the spectrotemporal statistics of those sounds (e.g. McDermott and Simoncelli 2011 ). — Dijk, Pim van, Physiology, Psychoacoustics and…, 3286
Categorization
- DEFINE: Defines auditory textures as sounds whose identity is carried by spectrotemporal statistics rather than deterministic elements — a key concept for the reader’s physiology/psychoacoustics focus.
- XREF: Links to the reader’s ongoing interest in how the auditory system abstracts statistical regularities, complementary to louder soundscape material from The Soundscape. — connects to: Auditory textures are stimuli whose only predictability lies in their statistical structure; Predictable sounds carry less information than unpredictable broadband signals
- corroborated: 2 independent passages support this claim — Dijk, Pim van, Physiology, Psychoacoustics and…, 3286