Robust multiple sound source localization in noisy environment by using a soundfield microphone

Sound source localization techniques are becoming popular as they provide an effective information for parameter coding and reconstruction of sound scene. A recent approach based on “single-source” zone detecting was proposed. However, the method is not robust in noisy environment due to its DOA estimation principle. To overcome this issue, a mixture enhancement processing based multiple sound source localization method using soundfield microphone is presented in this work. In addition, a dual threshold determination method of “single-source” zone is proposed as well. The combination of the two methods can effectively weaken the influence of noise on the localization. Thereafter, localization can be proceeded in these detected “single-source” zones by applying a peak searching approach to the normalized DOA estimation histogram. The evaluations reveal that the proposed method achieves a higher accuracy of DOA estimation compared with the existing techniques and it is robust in noisy environment.