Beschreibung:
<jats:p>A system was proposed by G. Ren et al. [158th Meeting of the ASA (2009)] to perform field acoustics measurements during ongoing music concerts. It addresses the long-standing problem of performing architectural acoustic measurements with an audience present. In this system, an adaptive time-frequency synthesis algorithm was employed to generate test signals with time-frequency energy distributions that conform to predicted “masked measurement vacancies” to ensure that the acoustic response of the test signals can be separated from the music and that the test signals are inaudible to the audience by taking advantage of auditory masking phenomena. In the present paper, approaches utilizing higher-level structure of music are exploited to reduce the number of conflicts between the test signals and the music. Three approaches are described: a music theory based probabilistic prediction algorithm, using prior knowledge of the score, and the third exploits the availability of previous recordings of the music being performed. These control algorithms are adapted to architectural acoustic measurement procedures conforming to ISO3382-1/2(2009). The proposed system achieves similar performance compare to conventional acoustic test procedures, which emit loud test signals in empty music halls.</jats:p>