Spatial Room Impulse Response Dataset: A Robot's Journey Through Coupled Rooms of a Reverberant University Building

Description

Technical info (English)

• The MetraLabs TORY robotic platform has a two-wheel drive and can therefore turn on the spot in addition to translation on the plane

• In previous studies, the positioning accuracy of the robotic platform in translation was determined to be 3.0 cm on average, with an average measurement accuracy of 2.2 cm. For the rotation, the positioning accuracy was determined to be 0.6° on average and the average measuring accuracy to be 0.7°.

• The recordings were made with a 7-piece microphone array. It consists of 6 electret microphones (called satelites) that are spherically distributed with a diameter of 100 mm around a Earthworks M30 measurement microphone.

• In relation to the measurment microphone the positions of the satellite microphones is as follows:

  Microphone	Audio Channel	x [mm]	y [mm]	z [mm]
  Satellite 1	1	50	0	0
  Satellite 2	2	0	-35	-35
  Satellite 3	3	0	-35	35
  Satellite 4	4	-50	0	0
  Satellite 5	5	0	35	35
  Satellite 6	6	0	35	-35
  Measurement Microphone	7	0	0	0

• The impulse responses where computed from exponential sine sweep recordings. The sine sweeps went from 40Hz to 22kHz in 10 seconds with additional 4 seconds of decay time afterwards.

• To reduce the occurrence of noise the measurements where conducted at night.

• The speakers used to excite the room were Geithain MO-2 for SSU, SMU, HS and HM and Geithain RL900A for SSL and SML.

• The SRIRs were computed from the sine sweep recordings using the Spatial Decomposition Method (SDM).

• This dataset is in SOFA 2.1 file format as described in AES69-2022. The files follow the convention SingleRoomSRIR Version 1.0.

Abstract (English)

A key challenge in auditory mixed reality is the plausible embedding of virtual objects in the real environment. Common approaches to determine the room acoustics are simplified geometry or wave-based room acoustic simulations, parametric description of acoustic room transfer functions, use of photos, videos or 3D scans to estimate transfer functions, or direct generation of spatial audio.

All these approaches result in a more or less correct approximation of the real room acoustics. An analytical and perceptual comparison between simulation and reality is therefore highly desirable. However, high-precision measurements at a large number of source-receiver positions in a room, which can be used as a reference, are complex to produce, require large amounts of data and are prone to error.

Therefore, the work presents a data set of spatial room impulse responses with a high spatial resolution of the interior spaces (hallway and staircases) of a building. The dataset covers coupled room acoustics, large and small distances between sources and receivers, multiple visible and occluded sound sources, sources in the hallway and at adjacent staircases on the same floor but also on the floor below.

Other (English)

Funding

This dataset was created as part of the projects "Multiparties" in the framework "KMU-innovativ: Interactive Technologies for Health and Quality of Life" of the Federal Ministry of Education and Research (BMBF), Germany and "ISOPERARE", funded by Reality Labs Research at Meta.

Notes (English)