运用神经网络推算音乐厅满场混响时间

Determination of the occupied reverberation time in concert halls by the use of neural network

混响时间是厅堂音质评价中的一个重要客观参量。本文提出了运用神经网络由空场混响时间推算满场混响时间的新方法。与其它方法比较,该方法有相当的预测精度。在此基础上,本文进一步分析了音乐厅中座椅类型和座椅数量对满场混响时间的影响,并对满场条件下座椅的吸声增量进行了讨论。

Reverberation time is one of the important objective parameters for evaluating acoustics in concert halls. A new method is proposed to determinate the occupied RT on the base of the unoccupied RT using neural network. Compared with other methods, it has the advantage of higher accuracy of prediction. Meanwhile, the impact of the type and the number of seats on the occupied RT is discussed and the sound absorption increment of the occupied seats investigated.