Audio Feature Analysis

Abstract

There are, in general, two types of audio features: the physical features and the perceptual features. Physical features refer to mathematical measurements computed directly from the sound wave, such as the energy function, the spectrum, the cepstral coefficients, the fundamental frequency, and so on. Perceptual features are subjective terms which are related to the perception of sounds by human beings, including loudness, brightness, pitch, timbre, rhythm, etc. In this research, we use the temporal curves of three kinds of short-time physical features, i.e. the energy function, the average zero-crossing rate, and the fundamental frequency, as well as the spectral peak tracks for the purpose of coarse-level audio segmentation and classification. While for the fine-level classification and retrieval, one of our most important tasks is to build physical and mathematical models for the perceptual audio features with which human beings distinguish the subtle differences among different classes of sounds. Currently, we consider two kinds of perceptual features in this work, i.e. timbre and rhythm.