Acoustic source classification using hyperset of fused voice biometric and spatial features

A non-transitory computer-readable medium having instructions stored thereon that are capable of causing or configuring a system to perform operations comprising:

• extracting, from input of multiple microphones, a hyperset of features of acoustic sources;
• wherein the hyperset of features comprises one or more spatial features of the acoustic sources and one or more voice biometric features of the acoustic sources;
• using the extracted hyperset of features to identify separable clusters associated with acoustic scenarios;
• wherein the acoustic scenarios comprise:
• a desired spatially non-moving talker;
• a desired spatially moving talker;
• an undesired spatially non-moving acoustic source;
• and
• an undesired spatially moving acoustic source;
• and
• classifying subsequent input of the multiple microphones as one of the acoustic scenarios using the hyperset of features.

An apparatus, comprising:

• a feature extractor that extracts a hyperset of features of acoustic sources from input of multiple microphones;
• wherein the hyperset of features comprises one or more spatial features of the acoustic sources and one or more voice biometric features of the acoustic sources;
• a clustering block that uses the extracted hyperset of features to identify separable clusters associated with acoustic scenarios;
• wherein the acoustic scenarios comprise:
• a desired spatially non-moving talker;
• a desired spatially moving talker;
• an undesired spatially non-moving acoustic source;
• and
• an undesired spatially moving acoustic source;
• and
• a classifier that classifies subsequent input of the multiple microphones as one of the acoustic scenarios using the hyperset of features.