Fig. 3From: Robustness of ad hoc microphone clustering using speaker embeddings: evaluation under realistic and challenging scenariosPerformance metrics (SIR, PESQ, STOI) showing the separation effectiveness of the cluster feature types (colours) and method (x-axis) for the first set of scenarios, where the sources are always sufficiently far apart. For this and the other scenarios, some audio examples are available at https://users.ugent.be/ sbkindt/EURASIP_ASN/Back to article page