eduzhai > Applied Sciences > Engineering >

FCN Approach for Dynamically Locating Multiple Speakers

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: In this paper, we present a deep neural network-based online multi-speakerlocalisation algorithm. Following the W-disjoint orthogonality principle in thespectral domain, each time-frequency (TF) bin is dominated by a single speaker,and hence by a single direction of arrival (DOA). A fully convolutional networkis trained with instantaneous spatial features to estimate the DOA for each TFbin. The high resolution classification enables the network to accurately andsimultaneously localize and track multiple speakers, both static and dynamic.Elaborated experimental study using both simulated and real-life recordings instatic and dynamic scenarios, confirms that the proposed algorithm outperformsboth classic and recent deep-learning-based algorithms.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×