README.TXT

Impulse response (IR) measurements were made in scale models of orchestra pits, as described in the paper which is available at http://www.jstage.jst.go.jp/article/ast/26/2/237/_pdf.

A small loudspeaker element was used as sound source and IRs were measured to two microphone positions for three different orchestra pit geometries:

	"No overhang"
	"1.5 m overhang"
	"3 m overhang"

Please note that the labels "3 m overhang" and "1.5 m overhang" refer to a full-scale equivalence of the orchestra pit. In the physical scale model, the overhangs were 5 times smaller: 0.3 m and 0.6 m.

A reciprocity technique was used by positioning the two microphones flush mounted in the floor of the orchestra pit, and by placing the loudspeaker in 296 positions along a line above the orchestra pit, in steps of 1 cm. Geometrical details can be found in the paper referred to above. The geometry is also described in the geometry files in the .cad format, as used by the software CATT Acoustic, www.catt.se. The .cad files are text files and easily readable. Please note that the source and receiver positions in the cad files do not include all positions, because the measurements used more source positions than can be handled by the CATT Acoustic software.

The impulse responses were measured with the WinMLS software (www.winmls.com) and are stored as binary WinMLS files at http://www.iet.ntnu.no/~svensson/software/. The WinMLS files can be imported directly into Matlab using the loadimp matlab function, which is also available at http://www.iet.ntnu.no/~svensson/software/.

The two microphones were 1/4-inch BrŸel&Kj¾r condenser microphones. The loudspeaker element was a small (1 inch SEAS) tweeter and its response was measured in angular steps of 10 degrees, from -50 degrees to +50 degrees (relative to on-axis radiation). One inverse filter for each channel/microphone has been derived. The inverse filter compensates for the non-flat loudspeaker response, as well as the slight difference in microphone sensitivity between the two channels/microphones.  The inverse filters are available at http://www.iet.ntnu.no/~svensson/software/.

In using these data for, e.g., comparing with simulations, the following steps are suggested to implement in Matlab code:

1. Load the two inverse filters into Matlab.

2. Choose one of the three geometry directories ("No overhang", "1.5m overhang", or "3m overhang"). Please note that the file names are:

	###_ch#*.wmb
	
	where the first ### go from 0 to 295 and represent the source positions
		  the second # is 1 or 2 and represent the microphone position
		  the last * is a or b or "", indicating the different pit geometries.

3. Load one IR after another using the loadimp function. 

4. Convolve each measured IR with the appropriate inverse filter, i.e., either ir_inv1 or ir_inv2.


Peter Svensson and Anders Lovstad

Email to corresponding author: svensson@iet.ntnu.no