Multi Room Transition Dataset

Introduction We present a dataset of room impulse response (RIR) measurements with a focus on covering a broad spectrum of room-acoustic conditions. A central motivation was to collect data from complex, real-world multi-room environments in which acoustic energy decay functions (EDFs) are characterized by multiple decay slopes. The dataset is accompanied by positional information and visual documentation in the form of 360˚ photographs. A total of 4032 room impulse responses were measured throughout three complex indoor environments using a KEMAR head and torso simulator, and a Zoom H3 ambisonics microphone array. Acoustic environments In each of the three environments considered in the dataset (hallways-lecturehall, offices, workshops), four loudspeakers were placed and remained stationary throughout all measurements. A mobile measurement platform was then moved along a predefined path that covered various measurement positions. At each position, the Room Impulse Responses (RIRs) between the mobile platform and each of the four loudspeakers were recorded. Mobile measurement platform The mobile recording platform consisted of a head and torso simulator (GRAS 45BB KEMAR), fitted with binaural microphones, and a four-channel tetrahedral microphone array (Zoom H3-VR), placed on top of the head simulator. Signals All measurements were conducted at a sampling rate of 48 kHz (24-bit), for ease-of-use, we provide the signals in the dataset in two formats: as a pickled pandas dataframe (mrtd_dataframe.pkl) and as SOFA files. ------------------------------------------------------------------------------------------------------------------------------------------------------------------- Publication A more detailled description of the dataset and a method for the blind estimation of Energy Decay Functions can be found in our IWAENC 2024 paper: Philipp Götz, Georg Götz, Nils Meyer-Kahlen, Kyung Yun Lee, Karolina Prawda, Emanuël A. P. Habets, and Sebastian J. Schlecht: A Multi-Room Transition Dataset for Blind Estimation of Energy Decay, IWAENC 2024, Aalborg, Denmark The accompanying code repository can be found at GitHub.