AES 東京コンベンション 2009
Paper Session A5

Two experiments and two analyses were performed in order to evaluate sounds of snare drums quantitatively as a preparative study for quantitative evaluation of drum sets. The experiments were performed with a set of stimuli generated by 10 snare drums. ``Method of Paired Comparison'' and ``Rating Scale Method'' were used to the experiments. ``Multi Dimensional Scaling(MDS)'' and ``Multiple Regression Analysis'' were used in the analyses. From these, connections between attribute ratings for each snare drum's sounds and the two-dimensional stimulus space based on psychological distance were obtained. Besides, from a supposition that the sounds of snare drums are consisted of high-and low-frequency parts separated at 2kHz that each of them mainly contain sounds of a snare and drum's shells respectively, ``Spectral Centroid'' and ``Sustain Time 60(ST60)'' for the two parts were calculated, and Correlation Coefficient between the physical measure and the two-dimensional stimulus space were calculated. A series of studies revealed that each dimensions achieved by MDS were related to ST60 (especially sustain of drum's shells) and Spectral Centroid below 2kHz (pitch of drum's shells).

When auditory attribute ratings are dependent on experimental context, incorporating task variables into prediction models should improve the success of those models. One such task variable is the way in which stimuli are blocked together into multiple comparison trials for collecting ratings. In this study, contextual dependence of physical predictors for two auditory attributes, apparent source width and sharpness, was examined by incorporated stimulus blocking factors into the prediction equations. The results showed that such an approach to prediction gave a much better fit for obtained sharpness ratings than did conventional physical measures. In contrast, ratings of auditory source width were found to be much less affected by context, perhaps because listeners rely upon a more absolute internal reference for spatial extent when making subjective judgments of width.

Loudspeaker array design employing M-S technique is a novel approach in sound reinforcement. The experimental design proposes a number of advantages over conventional L-R stereo system for applications in proscenium theaters and black box theaters. It is evident that a solid center sonic image and consistent stereo width can be enjoyed across a broader area in the auditorium especially with low aspect ratio when compared to L-R stereo system. This paper summarizes the attempts of M-S speaker array design, alignment procedures and signal processing applied in previous theatrical productions, and a speaker system for subjective testing is devised. Various stereo music materials are tested and the subjective test results are discussed.

A concert hall Sogakudo has adjustable ceiling panels. Acoustical changes due to the adjustable ceiling panels were investigated, and how those changes affect the recordings made in the venue. Impulse responses were recorded at seven settings of ceiling panels. Especially, recordings using the ``Omni-8'' array were made at reference position, 1m near, and 1m far from the loudspeaker. Subjective evaluations of 12 sound stimuli of four ceiling panel settings and three frontal microphone positions were done. Three music excerpts were convoluted with the 12 impulse response files resulting in 36 music stimuli. Three attributes were used for the evaluation. As a result, we got some information such as ceiling panel setting usually used for piano was rated higher in "Comfortableness" in all stimuli played solo.