Acoustic simultaneous localization and mapping (A-SLAM) of a moving microphone array and its surrounding speakers

Abstract

Acoustic scene mapping creates a representation of positions of audio sources such as talkers within the surrounding environment of a microphone array. By allowing the array to move, the acoustic scene can be explored in order to improve the map. Furthermore, the spatial diversity of the kinematic array allows for estimation of the source-sensor distance in scenarios where source directions of arrival are measured. As sound source localization is performed relative to the array position, mapping of acoustic sources requires knowledge of the absolute position of the microphone array in the room. If the array is moving, its absolute position is unknown in practice. Hence, Simultaneous Localization and Mapping (SLAM) is required in order to localize the microphone array position and map the surrounding sound sources. In realistic environments, microphone arrays receive a convolutive mixture of direct-path speech signals, noise and reflections due to reverberation. A key challenge of Acoustic SLAM (a-SLAM) is robustness against reverberant clutter measurements and missing source detections. This paper proposes a novel bearing-only a-SLAM approach using a Single-Cluster Probability Hypothesis Density filter. Results demonstrate convergence to accurate estimates of the array trajectory and source positions.

Localization of moving microphone arrays from moving sound sources for robot audition

“…To address the challenge of SLAM of moving sensor arrays and moving sources typical for acoustic sensors, we propose to utilize the a-SLAM approach in [6], [7]. In this framework, the map of sources in (1) is captured in a RFS, S t {s t,n } Nt n=1 , where each source position is considered as a random state variable and where the set cardinality, N t , itself is a random variable.…”

“…A new approach for SLAM using acoustic sensors recently proposed in [6], [7] addresses the challenges of clutter, missing detections, and SSL errors encountered when mapping sound sources from reverberant speech signals. By formulating the multi-source states as a RFS, surviving and newborn sources, clutter, and missing detections can be explicitly modelled and exploited for estimation of the source positions using a Probability Hypothesis Density (PHD) filter [8].…”

“…By formulating the multi-source states as a RFS, surviving and newborn sources, clutter, and missing detections can be explicitly modelled and exploited for estimation of the source positions using a Probability Hypothesis Density (PHD) filter [8]. Results in [6] demonstrate that the proposed a-SLAM framework successfully infers the sensor path as well as the range and positions of static sources from the bearing-only measurements.…”

“…In this paper we propose a novel extension of the a-SLAM approach in [6], [7] by incorporating dynamical models to capture the movement of sound sources typically encountered in practice. The resulting a-SLAM approach captures the movement of walking human speakers and models the timevarying positions of static speakers due to head and body rotations.…”

Under-modelled blind system identification for time delay estimation in reverberant environments

“…Reliable DOA estimation is important on applications such as speaker tracking [1,2], beamforming-based multichannel speech enhancement [3,4,5] and dereverberation [6,7].…”