Statistical method reduces complexity of musical instrument sound patterns
A Compact Statistical View of Musical-Instrument Directivity
Sound
Summary
Musical instruments produce sounds that spread in complicated ways which change with pitch. The authors introduce a new way to describe these sound patterns using fewer numbers while keeping important details. This helps compare different instruments and analyze their sound more easily. Their method shrinks large data files into much smaller summaries, making it practical for audio applications. They tested it on many orchestral instruments and found it works well.
What this means in practice
- •For audio software developers: Create compact but accurate models of instrument sound radiation for spatial audio rendering and sound simulations.
- •For musical instrument makers: Analyze and compare frequency-dependent sound projection patterns of instruments to assist in design and evaluation.
Authors
Gioele Greco, Mirco Pezzoli, Raffaele Malvermi, Gian Marco Ricci, Fabio Antonacci
Abstract
Directivity is a defining feature of musical-instrument radiation and is known to vary strongly with frequency. With the growing availability of measured directivity datasets and their use in sound-field simulation and spatial audio, there is a practical need for representations that are both compact and comparable across sources. This work addresses that need by proposing a statistically grounded description of frequency-dependent directivity that reduces complex radiation patterns to a small set of meaningful descriptors. The approach enables directivity patterns to be summarized in an interpretable way and compared consistently across frequency bands, even when the radiation is not well described by a single dominant direction. The resulting representation supports tasks such as analysis, categorization, and model building, providing a bridge between detailed measurements and higher-level descriptors useful for musical acoustics and audio rendering. Experiments on a corpus of 41 orchestral instruments show that the proposed method preserves the distance geometry well while reducing the full order-4, 30-band corpus from about 492~kB to about 100~kB in double-precision storage.