The Darkside 50 detector at Gran Sasso

In February 2013, I was able to arrange for David Monacchi and Fons Adriaensen to record the acoustics (reverberation or "impulse response") of the almost completed Darkside 50 detector for dark matter WIMPs (weakly interacting massive particles) constructed deep inside a mountain at the Gran Sasso laboratory, Italy. Recordings were made both inside the 4-metre diameter spherical detector (inside which is the argon-filled WIMP detector itself) and in the much larger enclosing chamber. The sphere has since been filled with liquid scintillator (to detect neutrons), while the chamber has been filled with water (to detect muons). This was therefore virtually a once-in-a-lifetime opportunity.

They used Ambisonic B-Format techniques to record the full 3D surround ("periphonic") impulse response, using a Soundfield ST450 microphone. The result is a set of high quality four-channel soundfiles containing first-order B-Format signals, which together encode the full 3D space including height information. They have generously made the files freely available under the Creative Commons licence.

For the outer chamber, recordings were taken from three microphone positions. In the confined space of the spherical detector the microphone was placed centrally, with three different positions used for the loudspeaker generating the stimulus sound.

This collection comprises twelve 4-channel B-Format soundfiles, sample rate 48Khz. Recorded to 24bit precision they show a clean decay all the way down to -140dB. They are in standard WAVEFORMATEXENSIBLE format, which all modern audio workstation software will support. They can also easily be converted to AMB format, e.g. using programs in the CDP Multi-Channel Toolkit. The archive also includes configuration files for users of Jconvolver.

Impulse responses of this kind are typically used as input to a convolution reverb process. While they can always be applied to a plain mono sound source, a more idiomatic way to use them would be to convolve with source material also in first-order B-format form, whether recorded directly or synthesised e.g. by using a B-Format panner. The B-Format output will then need to be decoded to the required speaker layout. Some suggestions regarding tools for B-Format convolution and processing are listed on the right. An especially appropriate use of these files would be in the sonification of particle physics data, e.g. as provided by the LHCsound project.

Richard Dobson April 2014