A Multimodal Approach to Blind Source Separation of Moving Sources

A novel multimodal approach is proposed to solve the problem of blind source separation (BSS) of moving sources. The challenge of BSS for moving sources is that the mixing filters are time varying; thus, the unmixing filters should also be time varying, which are difficult to calculate in real time. In the proposed approach, the visual modality is utilized to facilitate the separation for both stationary and moving sources. The movement of the sources is detected by a 3-D tracker based on video cameras. Positions and velocities of the sources are obtained from the 3-D tracker based on a Markov Chain Monte Carlo particle filter (MCMC-PF), which results in high sampling efficiency. The full BSS solution is formed by integrating a frequency domain blind source separation algorithm and beamforming: if the sources are identified as stationary for a certain minimum period, a frequency domain BSS algorithm is implemented with an initialization derived from the positions of the source signals. Once the sources are moving, a beamforming algorithm which requires no prior statistical knowledge is used to perform real time speech enhancement and provide separation of the sources. Experimental results confirm that by utilizing the visual modality, the proposed algorithm not only improves the performance of the BSS algorithm and mitigates the permutation problem for stationary sources, but also provides a good BSS performance for moving sources in a low reverberant environment.