Localization of sound sources in three dimensions is an essential component of immersive audio applications that use an object-based approach. Typically, sound sources are localized by triangulating direction of arrival (DOA) estimates from multiple, spatially distributed microphone arrays. We present an alternative approach operating on time-frequency representations of the steered response functions (SRF) obtained from spherical microphone array (SMA) signals. The method relies on the identification of bins that predominantly include a single source via the residual energy test (RENT) using a combined dictionary designed for multiple SMAs. DOAs of sources are then obtained by clustering individual DOA estimates at each time-frequency bin. The distances of sound sources are estimated via ternary search in the direction of the identified DOA vector. We show the utility of the proposed method under highly reverberant conditions by an example using three SMAs.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22525
(9742KB)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
