MEASURING IMPULSE RESPONSES IN A FULLY OCCUPIED CONCERT HALL USING A TAILOR-MADE ELECTRONIC MUSICAL COMPOSITION - PROOF OF CONCEPT

摘要

Since the beginning of concert hall acoustics research it is quite difficult or almost impossible to find or arrange an opportunity to do extensive measurements with audience present, shortly before the beginning of a symphonic concert or during a break. It is always a trade-off between time, type of signal, number of measurements and measurement quality, while the ISO 3382-1 standard ('Measurement of room acoustic parameters') demands a certain minimum amount of measurement positions and a certain minimum quality of measurement results (minimum decay ranges) depending on hall dimensions, number of seats, stage dimensions (presence of an orchestra pit) and parameter type. In the past there have been many successfull and unsuccessfull attempts to obtain room acoustical parameter values in occupied concert halls with and without orchestra using gun shots, balloon pops, noise bursts, etcetera. This type of signal is perceived as annoying and irritating (loud impulsive sounds and the smell of sulfur when using blanks) to both the audience and the orchestra. It disturbs the concentration of musicians and conductor before a performance, while the audience tends to react, producing additional sound. To circumvent this problem it is often attempted to use stop chords or loud impulsive fragments during a live performance of a symphonic orchestra. This was first done in 1935 by Meyer and Jordan during a symphonic concert in the occupied concert hall of the former 'Berliner Philharmonie' (Beethoven Coriolan-overture, Op. 62). To use this technique and reach a certain minimum acoustical quality of stop chords and impulsive sounds (time shape and spectrum) Jan Masseus (1913-1999) a Dutch composer and music pedagogue was contracted in 1973 to compose a symphonic piece of music for testing the acoustics of a fully occupied concert hall. He incorporated 'measurement signals' and the subsequent silences in a piece of symphonic music that was acceptable to the audience, which delivered better measurement results. However, the responses obtained from all above-mentioned measurements were mainly used to obtain an indicative value of the reverberation time. In a room acoustic study of Hidaka et al. in 2001 seven concert halls with and without audience could be measured using the ISO 3382 standard but 15 other halls with audience could only be measured using stop chords recorded during concerts with audience. It was found that the stop chord method has about twofold spreads to the ISO method. This means that very careful examinations are necessary when musical stop chords are used to obtain the reverberation time of a hall. In a study of Cox et al. it is described how to extract the room reverberation time from speech, using artificial neural networks. In his study, Kendrick describes the Maximum Likelihood Estimation (MLE) method. This statistical technique enables acoustical measurements in an occupied room, using only passively recorded speech and music signals.