Texturedness decision time for audio texturedness indicator

摘要

In this paper we analyze the texturedness decision time Td needed to evaluate the texturedness level of a given audio signal. We correlate the subjective values of Td with the objective time parameter Tobs of a recently proposed audio texturedness indicator [1]. As Tobs varies, speech texturedness measures are highly variant but got low texturedness values, whereas classic audio textures and academic noise got more stable and high texturedness values. A survey conducted on 56 subjects has shown that the very textured audio signals were quickly recognized and scored in a very short decision time, whereas speech signals decision time was higher. This subjective analysis suggests a rising time as the minimum time duration needed to make a texturedness score decision which must be included in the objective indicator.