eduzhai > Applied Sciences > Engineering >

Graph2Speak Improving Speaker Identification using Network Knowledge in Criminal Conversational Data

  • Save

... pages left unread,continue reading

Document pages: 8 pages

Abstract: Criminal investigations mostly rely on the collection of speechconversational data in order to identify speakers and build or enrich anexisting criminal network. Social network analysis tools are then applied toidentify the most central characters and the different communities within thenetwork. We introduce two candidate datasets for criminal conversational data,Crime Scene Investigation (CSI), a television show, and the ROXANNE simulateddata. We also introduce the metric of conversation accuracy in the context ofcriminal investigations. By re-ranking candidate speakers based on thefrequency of previous interactions, we improve the speaker identificationbaseline by 1.2 absolute (1.3 relative), and the conversation accuracy by2.6 absolute (3.4 relative) on CSI data, and by 1.1 absolute (1.2 relative), and 2 absolute (2.5 relative) respectively on the ROXANNEsimulated data.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...