Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing
Fig. 5 | EURASIP Journal on Audio, Speech, and Music Processing

Fig. 5

From: DOA-guided source separation with direction-based initialization and time annotations using complex angular central Gaussian mixture models

Fig. 5

DOA time annotations (DTAs) \(\beta ^{\text {dir}}_{k}(t)\) for J=1 speaker moving from φ=40∘ to 60∘. This results in D=3 discrete directions, and thus K=4 components. The function \({\mathcal {F}_{t}(j)}\) specifies, for each frame, which component corresponds to the j-th speaker. Note that DTAs and \({\mathcal {F}_{t}(j)}\) are determined solely by the DOA estimates. Nevertheless, the DTAs are unique, and thereby enable to enforce a fixed permutation

Back to article page