SPEAKER LOCALIZATION EXPLOITING SPATIAL-TEMPORAL INFORMATION

摘要: Determining the spatial position of a speaker finds growing interest in video conference scenario where automated camera steering and tracking are required. Speaker localization can be achieved with dual step approach. In preliminary stage microphone array is used to extract time difference arrival (TDOA) speech signal. These readings then by second for actual localization. Since trajectory must smooth, estimates close positions might improve current estimate. However, many methods, although exploiting information obtained dierent pairs, do not exploit this temporal information. contribution we present two schemes, which The first well known extended Kalman filter (EKF). recursive form Gauss method, denote Recursive (RG). Experimental study supports potential proposed methods.