6DoF SELD Dataset

6DoF SELD Dataset The 6DoF SELD Dataset is a SELD dataset for detecting and localizing sound events from the view of self-motion humans. Unlike conventional SELD datasets using fixed microphone arrays or wearable SELD datasets using HATS, sound events are recorded by a headphone-type device worn by a subject performing 6DoF self-motion (walking and looking around). The headphone-type device is equipped with an 18-channel microphone array and three motion tracking sensors. The motion tracking sensors allow the head's position and posture to be observed. The position and posture acquired by the motion tracking sensors can be time differentiated to simulate the observation of head motion by more practical sensors, such as a 6-axis inertial measurement unit (IMU). License: see the file named LICENSE.pdf Further information is available at [1] and Github: https://github.com/nttrd-mdlab/6dof-seld [1] Masahiro Yasuda, Shoichiro Saito, Akira Nakayama, Noboru Harada “6DoF SELD: Sound Event Localization and Detection Using Microphones and Motion Tracking Sensors on self-motioning human,” in IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP), 2024.